UNIVERSITY OF CALIFORNIA, IRVINE
Carbon MEMS from the Nanoscale to the Macroscale: Novel Fabrication Techniques and Applications in Electrochemistry
DISSERTATION
Submitted in partial satisfaction of the requirements for the degree of
DOCTOR OF PHILOSOPHY in Mechanical and Aerospace Engineering
by
Rabih Bachir Zaouk
Dissertation Committee: Professor Marc Madou, Chair Professor Donald Dabdub Professor Patrick Farmer
2008
© 2008 Rabih Bachir Zaouk
Dedication
To my parents, Randa and Bachir, my sisters Safa and Sarah, to Abir, to Fouad and Cynthia, to Juju and Khalouk, for their unconditional love to Malek, to Ra’fat, to all the Araneb, for showing me the true meaning of friendship
iii
Table of Contents Page List of Figures ................................................................................................................... vii List of Tables ................................................................................................................... xiii Acknowledgements.......................................................................................................... xiv Curriculum Vitae .............................................................................................................. xv Abstract of the Dissertation ........................................................................................... xxvi Chapter 1............................................................................................................................. 1 1.
Introduction................................................................................................................. 1 1.1. Micro ElectroMechanical Systems ..................................................................... 1 1.2. What is Carbon MEMS?..................................................................................... 3 1.2.1. Origins of C-MEMS ................................................................................... 4 1.2.2. Standard SU-8 Based C-MEMS Process .................................................... 5 1.3. Importance of Multiscale Carbon Structures .................................................... 17 1.4. The case for fractals in C-MEMS ..................................................................... 19
Chapter 2........................................................................................................................... 27 2.
Physically Modified C-MEMS Processes................................................................. 27 2.1. Rapid Freestanding SU-8 Photolithography Process and its Application in C MEMS ........................................................................................................................... 27 2.1.1. Introduction............................................................................................... 28 2.1.2. Materials and Methods.............................................................................. 30 2.1.3. Results and Discussion ............................................................................. 32 2.1.4. Conclusion ................................................................................................ 37 2.2. A Stretching Flow Polymer Deposition Technique for the Fabrication of Suspended Carbon Nanowires ...................................................................................... 38 2.2.1. Introduction............................................................................................... 39 2.2.2. Materials and Methods.............................................................................. 39 2.2.3. Results and Discussion ............................................................................. 43 2.2.4. Conclusion ................................................................................................ 54 iv
Chapter 3........................................................................................................................... 56 3.
Electrochemical Modifications to C-MEMS Techniques......................................... 56 3.1. PPY C-MEMS .................................................................................................. 58 3.1.1. Introduction............................................................................................... 58 3.1.2. High Overpotential Electrodeposition of PPY.......................................... 60 3.1.3. Templated Electrodeposition of PPy ........................................................ 63 3.1.4. Conclusion ................................................................................................ 69 3.2. Tools for Fractal Micro-Electrode Fabrication Using a Combination of Photolithography and Electrodeposition....................................................................... 69 3.2.1. Introduction............................................................................................... 69 3.2.2. Materials and Methods.............................................................................. 70 3.2.3. Results....................................................................................................... 71 3.2.4. Discussion ................................................................................................. 83 3.2.5. Conclusion ................................................................................................ 89
Chapter 4........................................................................................................................... 91 4.
Mixed Chemical and Physical Modifications to C-MEMS ...................................... 91 4.1. Porous C-MEMS............................................................................................... 91 4.2. Plasma Treatment.............................................................................................. 93 4.2.1. Conclusion ................................................................................................ 95 4.3. C-MEMS Mixed with Nanostructures .............................................................. 96 4.4. C-MEMS with in situ Grown Nanofibers....................................................... 103
Chapter 5......................................................................................................................... 105 5.
C-MEMS Fabricated Lithium Ion Anodes: Testing and Modeling ........................ 105 5.1.
Three
Dimensional
C-MEMS
Microbattery
Anodes
from
Pyrolyzed
Freestanding SU-8 Polymer Films: Fabrication Process and Testing of Lithium Ion Intercalation Properties ............................................................................................... 105 5.1.1. Introduction............................................................................................. 106 5.1.2. Materials and Methods............................................................................ 106 5.1.3. Results and Discussion ........................................................................... 115 5.1.4. Conclusion .............................................................................................. 147 5.2. Enhanced FEA Modeling Approach for 3D Microbatteries. .......................... 148 5.2.1. Introduction............................................................................................. 148 5.2.2. Method Specifics and Techniques .......................................................... 149 v
5.2.3.
Conclusion .............................................................................................. 164
Chapter 6......................................................................................................................... 165 6.
Other Applications of Carbon MEMS in Electrochemistry.................................... 165 6.1. Dielectrophotretic Trapping for Novel Photolithography Techniques ........... 165 6.1.1. Introduction............................................................................................. 165 6.1.2. Particle-Based Nanolithography ............................................................. 169 6.1.3. Particle – Substrate Compatibitlity ......................................................... 170 6.1.4. Nanomanipulation Issues ........................................................................ 174 6.1.5. Scaling Laws........................................................................................... 175 6.1.6. C-MEMS................................................................................................. 184 6.1.7. Proposed Lithography System and Validation ....................................... 186 6.1.8. Conclusion .............................................................................................. 193 6.2. Suggested Future Work................................................................................... 194
References....................................................................................................................... 195 Appendix......................................................................................................................... 202
vi
List of Figures Page Figure 1-1
SU-8 Based C-MEMS Process…………………………………………...6
Figure 1-2
Carbon Micro-Walls………………………………………………………7
Figure 1-3
Shrinkage in Carbon Micro-Walls………………………………………..8
Figure 1-4
Bad Adhesion of SU-8 to SiO2…………………………………………..10
Figure 1-5
Bent SU-8 Microstructures………………………………………………12
Figure 1-6
Wrinkled SU-8 Surface…………………………………………………..13
Figure 1-7
Two level SU-8…………………………………………………………..15
Figure 1-8
Fabricated Fractal C-MEMS……………………………………………..22
Figure 1-9
Schematic of Fractal C-MEMS…………………………………………..23
Figure 1-10
Koch Curves……………………………………………………………...24
Figure 1-11
Branching Fractal Tree…………………………………………………..25
Figure 2-1
SU-8 Layer Peeling Process……………………………………………...31
Figure 2-2
Bent Freestanding SU-8 Layer…………………………………………...34
Figure 2-3
Examples of Freestanding SU-8 Samples……………………………......35
Figure 2-4
Other Examples of Freestanding SU-8 Samples…………………………36
Figure 2-5
C-MEMS Carbon Thin sheets……………………………………………37
Figure 2-6
High Aspect Ratio Under-developed C-MEMS…………………………45
Figure 2-7
Stretching Flow Profile ………………………………………………….46 vii
Figure 2-8
SU-8 Nanowires Formation Sequence…………………………………...47
Figure 2-9
SU-8 Filaments Stretch from Post to Post……………………………….48
Figure 2-10
SU-8 Suspended Nanowire………………………………………………49
Figure 2-11
Array of Nanowires………………………………………………………50
Figure 2-12
Perspective View of an Array of Nanowires………………………….....51
Figure 2-13
Carbon Suspended Nanowire……………………………………………52
Figure 2-14
Nanowires Bending Posts………………………………………………..53
Figure 3-1
Pyrrole Monomer………………………………………………………...59
Figure 3-2
Overpotential Deposition of PPy………………………………………...61
Figure 3-3
PPy Cauliflower Structures………………………………………………62
Figure 3-4
Hydrogen Bubbles Templated PPy Deposition……………………….....64
Figure 3-5
Spherical PPy Shapes……………………………………………………64
Figure 3-6
PTFE Tape……………………………………………………………….65
Figure 3-7
PPy Forest from Templated Deposition in PTFE………………………..66
Figure 3-8
PPy Templated Deposition in PTFE……………………………………..67
Figure 3-9
Pyrolyzed PPy Structures………………………………………………..68
Figure 3-10
Bromide Crystal from Electrooxidation of CTAB………………………72
Figure 3-11
Bromide Crystal Roses on Top of C-MEMS Posts……………………...73
Figure 3-12
Bromide Crystal Sheets …………………………………………………74
Figure 3-13
CV of CTAB on Carbon…………………………………………………75
Figure 3-14
Chronoamperometry of CTAB on Carbon………………………………75 viii
Figure 3-15
CV of CTAB on Copper…………………………………………………76
Figure 3-16
CV of CTAB and PPy……………………………………………………76
Figure 3-17
CV of CTAB, PPy, and APS…………………………………………….77
Figure 3-18
EDS Results of Bromide Crystal on Carbon…………………………….77
Figure 3-19
EDS Results of Bromide Crystal on Copper……………………………..78
Figure 3-20
Bromide Crystal in the Presence of APS and Py………………………...78
Figure 3-21
Floral Deposition Pattern with CTAB in APS and Py…………………..79
Figure 3-22
Bristles …………………….…………………….………………………79
Figure 3-23
Long Range Bromide Crystal…………………….……………………...80
Figure 3-24
Short Range Bromide Crystals…………………….……………………..80
Figure 3-25
Evaporation of CTAB after Pyrolysis……………………………………81
Figure 3-26
C-MEMS Post with Bromide Crystal and PPy…………………………..81
Figure 3-27
Pyrolyzed C-MEMS Post with Bromide Crystal and PPy……………….82
Figure 4-1
Microporous C-MEMS…………………….……………………………92
Figure 4-2
Plasma Treated C-MEMS…………………….…………………………93
Figure 4-3
EIS of C-MEMS…………………….…………………………………...94
Figure 4-4
Morphology of CNF…………………….……………………………….97
Figure 4-5
Oxidized CNF on C-MEMS …………………….………………………98
Figure 4-6
SU-8 CNF Micropost …………………….…………………………….99
Figure 4-7
CNF covering C-MEMS Post…………………….………………….…100
Figure 4-8
Catalytic Growth of CNF…………………….…………………………101 ix
Figure 4-9
Cobalt C-MEMS Posts…………………….……………………………102
Figure 4-10
Unidentified Nanowires…………………….…………………………..103
Figure 4-11
CNT deposited on C-MEMS Posts……………………………………..104
Figure 5-1
Pyrolyzed SU-8 Films…………………….……………………………110
Figure 5-2
Cell Assembly…………………….………………………………….…113
Figure 5-3
Cell Assembly Connected…………………….………………………...114
Figure 5-4
Gravimetric Yield of Pyrolysis…………………….…………………...116
Figure 5-5
Rapid Pyrolysis Morphology…………………….……………………..119
Figure 5-6
Volumetric Yield of C-MEMS Pyrolysis………………………………121
Figure 5-7
Thickness of C-MEMS Carbon Sheets…………………………………123
Figure 5-8
Surface Shrinkage of C-MEMS SU-8…………………………………..123
Figure 5-9
Density of C-MEMS Sheets……………………………………………124
Figure 5-10
Density of Layers of SU-8……………………………………………...125
Figure 5-11
Charge Discharge Profile of C-MEMS Sheets…………………………126
Figure 5-12
Intercalation Capacity …………………….……………………………127
Figure 5-13
Specific Capacity …………………….………………………………...128
Figure 5-14
Volumetric Capacity…………………….……………………………...128
Figure 5-15
Irreversible Capacity…………………….……………………………...129
Figure 5-16
Capacity vs. Discharge Rate…………………….……………………...129
Figure 5-17
Volumetric Capacity at 3mA…………………….……………………..130
Figure 5-18
Capacity Loss at Higher Discharge…………………………………….131 x
Figure 5-19
EIS Nyquist from Zsimpwin…………………….……………………...132
Figure 5-20
Circuit Models…………………….……………………………………133
Figure 5-21
EIS Nyquist for Sheets…………………….……………………………133
Figure 5-22
Charge Transfer Resistance…………………….……………………....134
Figure 5-23
Exchange Current Density…………………….………………………..135
Figure 5-24
Change in Cdl with cycling…………………….………………………135
Figure 5-25
Diffusion Coefficient of lithium in C-MEMS sheets…………………..136
Figure 5-26
EIS Nyquist of Li metal…………………….…………………………..136
Figure 5-27
Randles Cell…………………….…………………….………………...150
Figure 5-28
Nyquist Plot from Randles modeling…………………………………...150
Figure 5-29
Nyquist Plot of Carbon Layer…………………….…………………….153
Figure 5-30
Electric Potential Distribution of a Microbattery………………………154
Figure 5-31
Current Density Distribution of a Modeled Microbattery……………...155
Figure 5-32
Current Density Profile at Top of Electrodes…………………………..156
Figure 5-33
Current Density Profile at Bottom of Electrodes………………………157
Figure 5-34
Isopotential Lines Distribution…………………….…………………...159
Figure 5-35
Electric Potential Distribution…………………….……………………159
Figure 5-36
Current Density Distribution…………………….……………………...160
Figure 6-1
Particle Based Nanolithography Concept………………………………170
Figure 6-2
Exposure from Single Phosphorescent Particle………………………...173
Figure 6-3
Exposure from Dragged Phosphorescent Particle……………………...173
Figure 6-4
Kodak Film Control Sample…………………………………………....174 xi
Figure 6-5
High Aspect Ratio C-MEMS Structures………………………………..186
Figure 6-6
Illustration of Particle Based Lithography using DEP Traps…………...188
Figure 6-7
Pyrolyzed C-MEMS Array…………………….……………………….190
Figure 6-8
C-MEMS Micropost…………………….……………………………...190
Figure 6-9
Dielectrophoretic Trapping of Polystyrene Beads……………………..190
Figure 6-10
Cluster of Polystyrene Beads Inside a Trap……………………………191
Figure 6-11
Electroosmotic Writing Mechanism……………………………………192
xii
List of Tables Page Table 1-1
Results of Fractal Modeling…………………………………………..…26
Table 2-1
SU-8 process parameters…………………….…………………………...26
Table 4-1
Electrochemical Parameters Extracted from EIS………………………...95
Table 5-1
SU-8 Process Parameters for Different Thicknesses…………………...108
Table 5-2
Pyrolysis Outcome Summary…………………….…………………….111
Table 5-3
Zsimpwin Results from EIS Analysis………………………………….134
xiii
Acknowledgements I would like to thank Professor Marc Madou, my mentor, my inspiration, and my friend, for supporting me throughout my Ph.D. journey. I would also like to thank Dr. Benjamin Park for being an exceptionally supporting labmate and friend. I thank Dr. Horacio Kido and Dr. Jim Zoval for the many discussions they’ve shared with me, which helped me along the ways of both academia and life. I would also like to thank NSF, CCAT, UC Discovery, and CMB who funded various parts of this work
xiv
Curriculum Vitae
Education
Ph.D. in Micro Electromechanical Systems (Mechanical and Aerospace Engineering department) from the University of California Irvine. (March 2008).
M.S. in Mechanical and Aerospace Engineering, University of California Irvine. (20002002).
B.E. in Mechanical Engineering, American University of Beirut. (1996–2000).
French Baccalaureate avec mention, Lycée Franco-Libanais Tripoli, Lebanon. (19821996).
Languages
Arabic, English, and French very well, G 3 level in German. Programming languages and Software.
xv
Languages: C++ , Fortran ,STEP 7 Siemens PLC programming.
CAD: Tanner’s Ledit MEMS design, Code V optics design, Autocad, Msc’s Patran FEA, FEMLAB, Ansys.
Others : MS office, Latex, Photoshop, Illustrator.
Experience
Chief Operating Officer, Carbon MicroBattery, LLC (May 2006- present)
Graduate Research Assistant with Professor Marc Madou, UCI (08/2002-present).
Teaching Assistant for PHY 3LB, the Basic physics II lab course in the Physics department at UCI. (Fall 2003 term).
Teaching Assistant for MAE10, introduction to engineering computations course in the mechanical engineering department at UCI. (Fall 2002 term).
Graduate research assistant at UCI Microsystems labs. (01/2001-08/2002).
xvi
Teaching assistant for ECE 70B, the network analysis course in the electrical engineering department at UCI. (Spring 2001 term).
Programmer for the experimental psychology department at UCI. (09/2000-01/2001).
Control and Process Engineer at Kamaplast plastic manufacturers. (01/2000-06/2000).
Internship in the Research and Development section of the SLV-DVS German welding institute in Munich. (06/99-10/99).
Awards and Accomplishments
Henry Samueli fellowship for graduate student in mechanical engineering (09/200109/2002)
Second prize at the UNESCO national exhibition for advanced scientific and industrial research, Lebanon (07/2000)
Developed a new shape of nails for HILTI int. used in the powder actuated welding technology, Germany (10/1999)
xvii
First price in the nationwide competition of " Maths sans frontieres" in high school as part of the winner class. Lebanon (1994) Community involvement
Member of the Faculty of Engineering and Architecture Student Representative Council for the academic year (1998-1999) AUB.
Organizer in the New Students Orientation Program at A.U.B. for two academic years (1997-1998) and (1998-1999).
Interests and activities
Rock climbing, ultimate Frisbee, tennis, ski, sailing, hiking, guitar playing, and salsa dancing.
xviii
PUBLICATIONS
Park, B.Y., R. Zaouk, C.L. Wang, and M.J. Madou, ”A case for fractal electrodes in electrochemical applications”. Journal of the Electrochemical Society, 2007. 154(2): p. P1-P5.
Wang, C., R. Zaouk, and M. Madou, “Local chemical vapor deposition of carbon nanofibers from photoresist”. Carbon, 2006. 44(14): p. 3073-3077
Park, B.Y., R. Zaouk, C. Wang, J. Zoval, and M.J. Madou, “Fractal carbon-mems electrodes: theory and preliminary fabrication”. ECS Transactions, 2006. 4(1, Microelectronics Technology and Devices): p. 83-92.
R. Zaouk, B. Y. Park, F. Galobardes, G. Turon, M. Madou, “Design and Characterization of 3D Carbon MEMS for Lithium Ion Microbatteries”, PowerMEMS 20006, Berkeley, CA.
Park, B.Y., R. Zaouk, C. Wang, and M.J. Madou, “Fractal carbon-MEMS architectures for 3D miniature power and sensor applications”. ECS Transactions, 2006. 1(20, ThreeDimensional Micro- and Nanoscale Battery Architectures): p. 1-11. xix
Rabih Zaouk, Benjamin Y. Park, and Marc J. Madou, “Photolithography as a Microfabrication Technique” Methods in Molecular Biology on Microfluidics, Humana Press, 2005
Rabih Zaouk, Benjamin Y. Park, Marc J. Madou, “Fabrication of PDMS Microfluidics Using SU-8 Molds,” Methods in Molecular Biology on Microfluidics, Humana Press, 2005
Benjamin Y. Park, Rabih Zaouk, and Marc J. Madou, “Microlithography for Microfabrication,” The Electrical Engineering Handbook, 3rd Edition, CRC Press, 2006
Benjamin Y. Park, Rabih Zaouk, Marc J. Madou, “Fabrication of Microelectrodes Using the Lift-off Technique,” Methods in Molecular Biology on Microfluidics, Humana Press, 2005 B. Y. Park, R. Zaouk, C. Wang, M. J. Madou, “Fractal Carbon-MEMS Architectures for 3D Miniature Power and Sensor Applications”, ECS conference proceedings, 2005
Benjamin Y Park, Rabih Zaouk, Marc Madou, “Validation of Lithography Based on the Controlled Movement of Light-emitting Particles,” SPIE Microlithography, Emerging Lithographic Technologies VIII, 2004
xx
V.K. Kayastha and Y.K. Yap (Michigan Technological University); C. Wang, R. Zaouk, L. Taherabadi, and M. Madou (University of California, Irvine) “Controlled Growth of Carbon Nanotubes for 3-D Lithium-Ion Microbatteries”, diamond conference 2005
C. Wang, R. Zaouk, L. Taherabadi, M. Madou, V. Kayastha, and Y. K. Yap, "3D Microbatteries with C-MEMS/CNFs and C-MEMS/CNTs Electrode Arrays," in 2004 MRS Fall Meetings, Boston, MA, Nov 29-Dec 3, 2004
C. Wang, R. Zaouk, L. Taherabadi, M. Madou, V. Kayastha, and Y. K. Yap, "CMEMS/CNTs Electrode Arrays for 3D Microbatteries," in 206th Meeting of the Electrochemical Society (2004 Joint International Meeting), October 3-8, Honolulu, Hawaii (2004), Symposium Q1, abstract 1492.
Chunlei Wang, Lili Taherabadi, Rabih Zaouk, Kartikeya Malladi, Marc Madou University of California, Irvine C-MEMS/NEMS: A Novel Technology for Graphite, Ni, and Si Nanoscale Material Formation ASME integrated nanosystems conference, Sep 2004, Pasadena.
xxi
MS THESIS
Literature Review and Preliminary Study on Feasibility of a MEMS Polarization Mode Dispersion Compensator, University of California, Irvine. Degree conferred on December 13, 2002.
TALKS
Rabih Zaouk, Marc Madou. "Carbon Microbatteries, Power on Chip", UCI MAE Seminar Series, Feb 2006
Rabih Zaouk, Marc Madou “From MEMS to NEMS with Carbon”. Nanoworld 2005, Los Angeles
Rabih Zaouk, Jitae Kim, Guangyao Jia, Kuo-Sheng Ma, Alia Marafie, Jim Zoval and Marc Madou ”CD-Based Nucleic Acid Analysis: Cell Lysis and Fast Hybridization Detection”, Nanotech 2003, Montreux
xxii
DISCLOSURES
UC Case No. 2004-054, "Advanced Non-conventional Lithography Techniques"
UCI Case No. 2005-047-1 “Method of Nanotexturing of Carbon Surfaces of HighAspect-Ratio Carbon-MEMS (C-MEMS) Devices”
UCI Case No. 2005-168-1 “Fabrication of Suspended Carbon Micro and Nanoscale Structures” UCI
Case
No.
2005-399-1
“Metal
Interconnects
for
Use
in
Carbon
Microelectromechanical Systems (C-MEMS)”
UCI Case No. xxxxx “Methods of obtaining porous C-MEMS electrodes”
UCI Case No. xxxxx “Methods for More Efficient Current Collection in Carbon Electrodes”
UCI Case No. xxxxx “Methods for Growing Nanofibers/Nanotubes on High Aspect Ratio Carbon Microstructures”.
xxiii
UCI Case No. xxxxx.” Method for obtaining fractal structures by combining self assembly and electrodeposition techniques”
UCI Case No. xxxxx. “Method of fabrication of freestanding SU-8 MEMS structures using a disposable polymer substrate”
UCI Case No. xxxxx..”A stretching flow based microfabrication technique for suspended SU-8 and Carbon nanowires”
FILED PATENTS
“Fabrication of high aspect ratio C-MEMS architectures”. 2005, (The Regents of the University of California, USA). Application: WO p. 32 pp.
“Surface and composition enhancements to high aspect ratio C-MEMS”. 2006, (The Regents of the University of California, USA). Application: US p. 19 pp.
xxiv
PROJECTS
Investigation of 3D lithography for use in medical electronics (in collaboration with Medconx).
Microjet Nozzle Design for Commercial Applications with Dr. S. Kassegne
Design and development of the CMB/UCI C-MEMS Lithium-ion battery
Fabry Perrot Array Project with Professor Richard Nelson
MISCELLANEOUS
Referee for Sensors and Actuators B: Chemical
Referee for IEEE Sensors
Nominated for “Best Teaching Assistant of the Year Award“, UCI 2002-2003 xxv
Abstract of the Dissertation Carbon MEMS from the Nanoscale to the Macroscale: Novel Fabrication Techniques and Applications in Electrochemistry By Rabih Bachir Zaouk Doctor of Philosophy in Mechanical and Aerospace Engineering University of California, Irvine, 2008 Chancellor’s Professor Marc Madou, Chair
Micro electromechanical systems (MEMS) have strongly impacted our way of life in the last two decades. From accelerometers and gyroscopes that ensure your driving safety, to inkjet printer cartridges that transpose your ideas onto paper, to micromirrors that enable your small projectors. MEMS have become more and more ubiquitous. Silicon, the material on which the semiconductor industry based its revolution, has so far been the material of choice for MEMS. While silicon is a great platform for constructing electronics, it is less than ideal for applications that involve electrodes exposed to aggressive liquid and gaseous environments. Carbon is one of the most commonly used materials when it comes to electrochemical applications, it is therefore the best candidate
xxvi
to carry over the trend of miniaturization in arenas such as smart chemical sensing, biological microdevices, miniature power, etc. Recent advances in engineering nanoscale structures show great promise towards delivering higher performance sensors, detectors, transistors, displays, etc. In order to leverage the power of nanostructures in general, new manufacturing processes that can bridge between the nanoscale and the macroscale are needed. Such integrated fabrication methods are essential in enabling the transfer of the advantages boasted by nanostructures from the research labs towards mass manufacturing. The present work starts by introducing the basic photolithography technique that has been used so far to fabricate Carbon MEMS (C-MEMS). Several novel techniques stemming for the original process are then described in details and lithium-ion microbattery anodes are presented as an example application of these novel fabrication methods. These Carbon MEMS anodes are characterized through a combination of cyclic voltammetry and electrochemical impedance spectroscopy (EIS). A new finite element analysis (FEA) technique is then proposed to more accurately model the current density distributions of 3Dimensional C-MEMS batteries without the need for very complex multiphysics modeling. The results show that variations in current density distributions, previously reported in the literature, were exaggerated. The work then moves to describe two other novel multiscale C-MEMS fabrication techniques that attempt to bridge the gap between macro and nano scales. The first involves the combination of electrochemistry and photolithography in order to achieve fractal like structures made entirely of carbon. The second uses a solution based xxvii
deposition technique that yields fine submicron glassy carbon wires suspended between microposts hundreds of microns apart. Although previously reported in the literature, the exact fabrication mechanism of these suspended nanowires was originally attributed to the wrong mechanism. The two phase flow deposition mechanism (i.e. stretching flow) is demonstrated and the original interpretation mechanism is rebutted. It is then shown how Carbon MEMS structures can be used in a multitude of applications e.g. dielectrophoresis, selective trapping, particle separation and manipulation.
xxviii
Chapter 1
1.Introduction
1.1.
Micro ElectroMechanical Systems
Micro ElectroMechanical Systems (MEMS), or the science of miniaturization, refers to a class of devices that have at least one of their dimensions in the micrometer range. While Integrated Circuit (IC) devices only involve electrical components (transistors, diodes, capacitors, etc.), MEMS devices take advantage of a wide range of other phenomena from mechanical to biological (BioMEMS). The materials and fabrication methods used in MEMS are much more varied than those used for IC fabrication (where one deals principally with silicon, oxides, and metals patterned using photolithography). In contrast to the IC industry where the devices are carefully packaged and protected from the environment, MEMS devices, such as pressure or glucose sensors, often must have 1
surfaces that are directly exposed to the environment they are sensing in. Examples of successfully commercialized mechanical devices are: accelerometers, gyroscopes, tilt meters, membrane pressure sensors, micro mirrors, optical MEMS switches, and inkjet print heads. Commercial BioMEMS devices include glucose sensors, lab-on-chip systems, and DNA arrays. Because of the irreversible chemical reactions involved and contamination considerations, BioMEMS devices tend to be disposable.
Silicon has been the platform of choice for most of the aforementioned applications. Most of the equipment used for MEMS is adapted from the IC industry which operates almost exclusively on silicon wafers. The growing MEMS industry has recently commanded its own fabrication tools and made it possible to move away from the costly choice of silicon as the exclusive substrate. Availability of equipment is not the only reason why MEMS are being fabricated with non silicon materials. The harsh environments that MEMS are exposed to require a variety of material properties that silicon does not possess. Inexpensive Polymers have often replaced silicon for disposable BioMEMS applications. Carbon can substitute for silicon for most other MEMS electrochemical applications. Though no commercial application has yet reached the market, MEMS that use carbon as a material have been used in a variety of research applications. The superiority of carbon can be summarized in the following points: -
Great resistance to highly corrosive environments
-
Very wide electrochemical stability window in aqueous solutions [1, 2]
2
-
Forms a large variety of functional groups upon treatment making it a convenient surface for binding biological agents [3]
-
Biocompatible
-
Electrically conductive
-
Relatively inexpensive
-
Carbon, as an element, forms polymers very easily that present a infinite choice of precursor materials that yield different physical and chemical properties
Carbon MEMS attempts to leverage the advantages of carbon within the miniaturized world.
1.2.
What is Carbon MEMS?
Carbon MEMS (C-MEMS) is simply defined as MEMS with functional elements made out of carbon. To qualify as MEMS, these systems need to have at least one of their dimensions in the micrometer range. Since micromachining carbon is difficult, the most common method of obtaining a C-MEMS structure is by first micromachining a polymer precursor (photoresist or other plastic) and then pyrolyzing at around 900oC in an nonreactive environment (inert or vacuum), thus transforming the polymer into carbon.
3
1.2.1.
Origins of C-MEMS
C-MEMS originated thousands of years ago when the first attempts at making charcoal succeeded. More recently, soft lithography techniques have been utilized to fabricate resonators and mechanical structures of various configurations [4]. The resulting structures were mostly planar with a small height to width ratio, from hereon referred to as aspect ratio. Higher aspect ratio microstructures were only obtained a few years later when regular lithography on an exceptionally transparent photoresist (SU-8) yielded carbon microstructures with an aspect ratio larger than 10 [5]. The work on SU-8 performed in M.J. Madou’s labs at UC Irvine’s forms the basis on which this dissertation is structured. The principal fabrication process is described in the following section. The high aspect ratio that SU-8 based C-MEMS exhibit received a lot of interest in various research applications which will be discussed in this work. It is sometimes easy to confuse the C-MEMS nomenclature with SU-8 based C-MEMS because the majority of the work on C-MEMS has so far used SU-8 as a precursor material. This dissertation will try to adhere to the more general definition of C-MEMS and will make distinctions to differentiate between the various microfabrication techniques that can be used to obtain C-MEMS devices.
4
1.2.2.
Standard SU-8 Based C-MEMS Process
SU-8 is the commercial name used by MicroChem©, MA, to refer to an epoxy novolac based negative photoresist that provides an exceptional contrast in the near UV range (350-400 nm)[6]. The resin, commonly referred to as Epon®, is a Shell Chemicals product that is very commonly used in the automobile industry for making durable interior parts and coatings. SU-8 has been called the poor man’s LIGA, because it was first utilized as an alternate method for obtaining very high aspect ratio structures required for electroplating through a mold typically achievable with X-ray lithography. SU-8 Based C-MEMS refers to a C-MEMS device that is obtained by pyrolyzing a microstructure made out of SU-8. The general description of the process is shown in Figure 1-1.
5
Photoresist
Substrate
Substrate
a) Photoresist spin coating
c) Development of a micro-pattern by removal of unexposed photoresist
Photomask
900 0 C
Photoresist
Substrate
Substrate
d) Pyrolysis at around 900 0C under a non-reactive atmosphere Figure 1-1: Schematic description of SU-8 based C-MEMS process. b) Ultraviolet exposure through a photomask
The process starts with spin coating of a layer of SU-8 (also known as Nano SU-8) onto a substrate (e.g. silicon wafer) to form a uniform photoresist layer anywhere from a few 100 nanometers to a few millimeters in thickness, depending on the spin regime and the solvent content of the specific type of resist used. A softbake is then used to remove the solvent from the photoresist prior to exposure to near UV light through a prefabricated photomask. UV exposure triggers a photo-acid generator that is amplified in a post exposure bake step that crosslinks the exposed SU-8 and makes it resistant to development. The sample is then developed in a developer solution to reveal the micropattern shown in Figure 1-1c. The sample is then subjected to pyrolysis, which consists of heating inside a furnace to around 900oC in a non reactive environment for 6
approximately an hour. Pyrolysis transforms the epoxy into a glassy carbon material with an amorphous structure. A reduction in volume occurs during pyrolysis with a yield of around 25% by mass. The base of the structure (below 10 μm from the substrate) being closely adhered to the substrate, does not deform noticeably, while the top of the structure shrinks considerably. The resulting carbon microstructures are a close replica of the initial photoresist pattern with a quasi isometric shrinkage (see Figure 1-2).
Figure 1-2: SEM image of carbon micro-walls approximately 20 μm wide and 150 μm high. Although the SU-8 structures had perfectly straight faces, the carbon structures show evidence of non-isometric shrinkage effects 7
In fact the shrinkage appears to be isometric in areas away from the interface between SU-8 and Substrate. The curved portion of the face is limited to the first 10 μm. Notice that the deformation is more noticeable in the direction of the largest in plane dimension as seen in Figure 1-3.
Figure 1-3: SEM image of the top of a C-MEMS wall. The footprint corresponding to the SU-8 structure is 95 μm long. The remaining carbon structure is 70 μm long. The shrinkage caused by pyrolysis in the length direction is 26% between the original SU-8 and the final carbon structure. If generalized to all dimensions, this shrinkage 8
would result in approximately 40% volumetric yield from the pyrolysis process (to be contrasted with the 25% mass yield). The details of SU-8 based C-MEMS processing are worth exposing in a detailed fashion, because they can teach us the extent to which we can push the fabrication process to our advantage. The following paragraphs will expose in a handbook style, the sequence of processes involved in fabricating C-MEMS using SU-8.
Choice of Substrate: The most commonly used substrate was a 500 μm thick prime Silicon wafer with a 500 nm layer of Thermally-grown Silicon oxide on the polished side. Adhesion of the final carbon structures is strongly dependent on the adhesion of the precursor SU-8 on the surface of the wafer. The bad quality of a batch of silicon dioxide was capable of bringing down the yield (C-MEMS not cracking) from around 100% to 20% (see Figure 1-4). SU-8 seems to adhere better to bare silicon than to a silicon dioxide surface. Care should be taken when handling the wafer. Simply touching the surface of the wafer with nitrile gloves can make SU-8 recede from the affected surface right after spin coating. Covering SU-8 with metals and subsequently pyrolyzing is a challenging process. It was found that SU-8 adheres well to nickel. Pyrolyzing SU-8 on top of most metals thinner than 1 μm always resulted in a breakdown of the carbon layer as well as the oxidation of the underlying metal.
9
Figure 1-4: Optical microscopy image of SU-8 microstructures on a bad SiO2 wafer. The image was taken right after the development process. All lines are supposed to be straight in a diagonal fashion. The space between neighboring circles is almost 150 μm. The detached lines self assembled in a periodic sinusoidal shape. Spin Coating: an excess of SU-8 is dropped on the surface of a wafer. The wafer is then typically spun at 500 rpm for 10 seconds, then at 3000rpm for 30 seconds. A uniform wafer coating is essential for the repeatability and reliability of the experiment at hand. It is always preferable to spin coat full wafers instead of coating smaller chips of wafers in order to avoid the non uniformity dictated by the edge effects in smaller samples. SU-8 tends to accumulate at the edge of chips and create edge beads. The effect of edge beads 10
is tolerable in full wafers, but in smaller samples, it can lead to a huge variation in photoresist thickness. The maximum recommended thickness of a continuous layer of SU-8 on silicon dioxide is 25 μm. Thicker layers would typically detach from the surface after pyrolysis. For a patterned layer with largest features smaller than 500 μm, thicknesses of SU-8 up to 800 μm have been achieved without peeling or cracking.
Soft Bake: Nano SU-8 is made of epoxy resin mixed in a solvent named Gamma Butyrolactone (GBL). The percentage of GBL varies from 60% for SU-8 2, down to around 25% for SU-8 100 [7]. A higher percentage of GBL makes the photoresist less viscous. The purpose of the softbake is to evaporate GBL and leave a solid SU-8 layer that is rigid enough to tolerate the harshness of contact photolithography. The baking is usually done on a hot plate in a two step process, first at 65oC then at 95oC. The two step process prevents heat shocking the sample and yields better SU-8 adhesion on the substrate (the 65oC step can be bypassed if adhesion is not very crucial). Thick layers of SU-8 (above 100 μm) can easily flow from one side of the wafer to another regardless of whether the spin coating was done properly or not. Carefully leveling the hot plate is strongly recommended for coatings thicker than 100 μm. The typical bake times quoted from the datasheet [6] provided by MicroChem© are around 20% shorter than the ones needed for a satisfactory result. Another way of assessing the proper baking time of a specific sample is to probe the side of the wafer with tweezers and check for a reasonable strength. If the coating is still very flexible, then more bake time is required. Excessive soft bake (beyond a couple of hours for layers thicker than 50 μm) can lead to cracking 11
and/or unintentional crosslinking of unexposed SU-8. When contact photolithography is used as an exposure method, a short soft bake can have catastrophic results. The SU-8 layer can easily warm up and flow to the point where it would stick to the photomask and damage it. A very flexible layer can also deform under the pressure of the contact and yield curved structure when the pressure is later removed as seen in Figure 1-5
Figure 1-5: SEM image of a 700 μm high SU-8 posts displaying a curvature that indicated bending during contact UV exposure. The thick layer was not baked long enough and SU-8 deformed from the heat and pressure of the exposure processing step. For extremely thick layers, a short softbake can lead to the formation of waves at the surface of the SU-8 during the post exposure bake due to different crosslinking levels at different thicknesses as seen in Figure 1-6. During softbake, the top layer is dried of its solvent content the quickest. The lower layers take longer to dry out. If solvent remains in those layers, it negatively affects both the exposure and the amplification process leading to a much less crosslinked layer. The lightly crosslinked layers at the bottom are sometimes not strong enough to sustain the shrinkage in the highly crosslinked top layer. 12
This problem is prominent when the mask exposes a large area with no microfeatures to release the accumulated sheet stress building within the thick layer of SU-8 as was for the sample in Figure 1-6. It is essential to allow sufficient cool down time after the soft bake before exposing the photoresist (at least ten minutes). Not doing so can negatively affect the minimum feature size of your micropattern and damage your photomask by gluing SU-8 on it during long exposures of thick layers (above 100 μm). This can be avoided by using intermittent exposures and dividing the exposure dose into smaller exposures whereby an SU-8 layer is allowed the time to dissipate the heat to the environment and not reaching a temperature that makes allow it to flow.
Figure 1-6: Photograph of a top wrinkled surface of a 1 mm thick layer of SU-8 that was soft baked for shorter than required. The photograph was take after the post exposure bake step
13
Exposure: during this step near UV light from mercury lamp is used to expose the SU-8 resist. UV radiation is absorbed by a Photo Acid Generator (PAG) creating a latent image of the microstructures that need to be formed. The PAG then forms a catalyst that chemically amplifies the effect of a single photo event in a 1.5-20 nm radius [8]. The quantum yield of such an efficient process can be as high as a 100. The concept was originally conceived in 1973 but was later developed by IBM in the 1980’s. The PAG in the case of SU-8 is an SbF6- salt mixed in with a ratio of 1.9-3.5% by mass, depending on the type of SU-8 [7]. While a typical exposure dose is around 400 mJ for a 100 micron layer, a very thin top layer (see Fig. 1-7) can be exposed with as small as 12mJ (as calculated from UV meter that integrates power from 350-430 nm coming out of a mercury lamp). Grayscale SU-8 construction can be achieved by varying the small exposure dose whether using UV or E-beam lithography [9]. Several levels of SU-8 structures can be stacked on top of each other without the need for meticulous alignment.
14
Figure 1-7: SEM image of an approximately 1μm thick layer of SU-8 sitting on top of an array of microposts. The top layer was exposed with 12 mJ of UV light. Post Exposure Bake: This is when the catalyst that is generated by the PAG is thermally activated to induce crosslinking of SU-8 epoxy. The post exposure bake (PEB) is usually done in a two steps, wherein a 65oC bake is followed by a 95oC bake. Similarly to a soft bake, a very long PEB can crosslink unexposed photoresist. It can also make the SU-8 more brittle and induce cracking. Development: SU-8 developer is a strong solvent known as Propylene Glycol Methyl Ether Acetate (PGMEA). SU-8 resin is highly soluble in PGMEA. Non crosslinked SU-8 in areas that were not exposed to UV light is quickly removed by a simple immersion in PGMEA helped along with mild stirring. SU-8 developer is reusable but its developing 15
power is minimized. PGMEA solution is later rinsed away with Isopropanol, which dissolves SU-8 to a much smaller extent than PGMEA but has much lower surface tension which helps avoid stiction effects. If the development process is not complete, a white precipitate consisting of quasi solid SU-8 resin appears. Further development needs to be done until all traces of white residue disappear. It’s easy to see that if the same developer solution is constantly reused, it can be impossible to get rid of the white residue (SU-8 resin) no matter how many times the rinsing process is repeated. The reuse of developer solution was primarily driven by the high cost of the product marketed as SU-8 developer. It was realized that generic solution of PGMEA purchased from Sigma Aldrich was five times cheaper and performed as good as the SU-8 developer sold by MicroChem©, MA (at least for feature sizes larger than 20 μm, development of smaller feature was not attempted)
Pyrolysis Process: Wang et al. [5] describe the general pyrolysis process as the following: -
A two step pyrolysis process in flow-through furnace using a quartz tube
-
Sample post baking for 40 min at 300oC under Nitrogen gas (N2).
-
Sample heated up to 900oC under N2 at around 10oC/min.
-
Sample maintained at 900oC under forming gas (95%N2, 5%H2) for 1 hour
-
Sample left to cool to room temperature under N2 for around 10 hours.
This process was successful in producing a good yield of C-MEMS when used in conjunction with SU-8. It is by no means the only way to produce C-MEMS from SU-8 16
as several variations of this process will be described later in this dissertation (e.g. forming gas is not necessary, and the two step heating process is not a must).
1.3.
Importance of Multiscale Carbon Structures
Since the discovery of buckyballs by Smalley et al.[10] and carbon nanotubes by Iijima[11], the whole world has turned its attention to the fascinating properties of these carbon singularities of nature. Nanotubes, nanoparticles, nanofibers, and nanoshells of various materials suddenly took center stage in science, as the materials that will bring about the next technological revolution. “Nano” as a buzz word became a selling point in a variety of industries: cosmetics, consumer electronics, clothing, etc. Many of these industries even started re-branding standard chemistry as nanotechnology. Despite the great promise that nanotechnology holds, So far, the real life impacts and the economic impacts have been rather “nanosized”. The reason behind this apparent infertility is that most of the projections that support the optimistic outlook for the future of nanoparticles are based on data that has been reported in the lab, in very specific environment, under very strict conditions, often on a single isolated nanoparticle or nanotube. Adoption of a technology often requires a fabrication process that is automated and scalable to mass manufacturability. It is this gap that Multiscale C-MEMS (MC-MEMS) hopes to bridge. The goal of this work is to investigate methods and techniques that can bridge the gap between the macro world in which we live in and the nano world where structures with 17
fascinating properties exist. MC-MEMS are defined as fabrication process that results in carbon structures that have features in several orders of the dimensional scale. MCMEMS is in a sense inclusive of the general concept of C-MEMS because often CMEMS is a carbon microstructure sitting on a substrate that is macrosized (the silicon wafer). The more granular control over the different scale levels is achieved, the closer to a biomimetic structures we can get. MC-MEMS will be a first step towards achieving fractal C-MEMS with carbon. This dissertation will describe several novel microfabrication techniques that can be categorized under MC-MEMS, and therefore serve as base stone for fractal C-MEMS construction. Although it is possible to obtain MC-MEMS using very simple fabrication methods (e.g. porous carbon), this work will focus on techniques that can for and MC-MEMS in addressable fashion, allowing it to integrated alongside other manufacturing processes. The difference between MC-MEMS and fractal C-MEMS is that fractal C-MEMS, to be faithful to its nomenclature, needs to exhibit self similarity [12]. Self similarity is an adjective used to describe objects that look the same when seen under different magnification. A mountain is often seen as a fractal because the grains of sand that constitute it resemble the general shape of the hill they form, which again looks like a miniature version of the larger mountain. Other examples of fractals in nature are fern trees, river systems, vascular systems, lungs, etc. While MC-MEMS are not fractals in the formal definition of the word, they can exhibit fractal like properties such as the ones described in section 1.4.
18
By modifying the original C-MEMS process, additional fabrication control over different scale levels can be achieved. Although it would’ve been nice to classify the novel microfabrication techniques presented in this dissertation under Nano, Micro, Macro CMEMS, many of the MC-MEMS processes span those different scales. The organization of the different techniques will thus be based on the type of modification to the initial CMEMS process required to obtain the new method (Physical, electrochemical, or a combination of chemical and physical).
1.4.
The case for fractals in C-MEMS
Since the identification of carbon nanotubes by Iijima [11], researchers have been struggling to integrate them within smart useful systems in an easy and efficient manner lured by the promises of better performing devices that the special characteristics of carbon nanotubes (CNT) can provide. In the decade that followed this invention, CNTs were being tested in every single application imaginable from structural composites to single molecule detection schemes. It was even visualized that CNTs would be the enabling technologies for space exploration scenarios involving a space elevator [13]. The recent years have seen widespread and growing interest in the fabrication and characterization of electrodes and electrode arrays with nanometer dimensions driven by the novel applications of such ultra-small devices and their unique electrochemical properties.
19
Carbon is a versatile electrode material that can undergo various chemical and electrochemical modifications to produce flexible surfaces ranging from purely inert to highly active surfaces. Nanometer-sized carbon electrodes could be of great significance in both fundamental and applied electrochemistry. For example, in miniaturized sensor arrays, it is the fact that its signal-to noise ratio will be increased by using ultra small electrodes. However, a lot of effects at the nanoscale, such as: mass transport, overall sensitivity, the number of molecules that can accumulate on a sensor, should be widely investigated in details. It will be very important from the fundamental research point of view to study the electrochemical properties and performance of these nanoelectrodes, and the local charge and mass transfer that occur at such solid-liquid interfaces and to understand the various new electrochemical properties of nanoelectrodes with dimensions near or even smaller than the diffusion length of chemical species in solution. Carbon electrodes modified with carbon nanotubes have demonstrated outstanding performance and excellent properties when used in electrochemical setups. When used for the potentiometric detection of morphine, the detection limit is 10 times better than the standard glassy carbon electrode [14]. Simultaneous detection of catechol and hydroquinone can be done using CNT modified carbon electrodes which is impossible to do with a regular glassy carbon electrode because the oxidation peaks of the two different chemical species are too close to be resolved [15]. Excellent electrocatalytic activity is also reported when the same type of electrodes are used in a glucose biosensors [16] and for the detection of neurotransmitters [17]. Further more, CNTs is a very promising material candidate in Li-ion battery applications. It has been reported that single-walled 20
carbon nanotubes can reversibly intercalate Li up to a rate of Li2.7C6 after applying an appropriate ball-milling treatment [18]. This storage capacity is higher than the so-called graphite intercalation compound (GIC), LiC6 in graphite (widely used as anode electrode material in commercial Li-ion battery). By incorporating CNTs in the electrodes it is possible to increase power density and decrease battery charge/discharge rates, while maintaining a high overall battery capacity.
Fractal or fractal-like C-MEMS design aims primordially at increasing the surface area of our electrodes in a manner that optimizes the performance of electrochemical devices. Alterations of every fractal order manufacturing step and correlating the change with the devices` performance (e.g. maximum power density for Li ion batteries) will allow us unprecedented control over the desirable parameters we want to optimize (e.g. signal to noise ratio in microsensors etc.)
In this work, we will fabricate and characterize carbon-based microelectrodes that exhibit a Nano/Micro fractal geometry by initial design. In contrast with the usual trend of, first fabricating the carbon nanostructures (tubes, fibers, etc.), and then lithographically defining an electrode at the convenient location on the substrate, Our novel methods integrate the fabrication of the Micro and the Nano using the same process step thus bridging that gap that separates these two scales. This bridging is a very important step in the direction of realizing mass-fabricated CNT-based microdevices which in the recent years have shown great promise for enhanced performance in a variety of applications 21
that include, but are not restricted to: nanoelectrochemistry, Nanoelectronics, field emission, chemical and physical sensing, biosensing, and composites.
Figure 1-8 shows an example of microstructure that have already been fabricated in the Madou group. The importance of an integrated Nano/Micro approach cannot be underestimated because it allows future viability of such a process in the face of constantly stringent hurdles that stand in between a conceptual fabrication process and its commercial adoption in a fabrication chain.
Figure 1-8: Examples of fractal microstructure that are achievable by combining the Carbon MEMS fabrication process with different nanofabrication methods to yield a fractal nano/micro carbon electrode.
The fractal configuration is embodied in the self similarity of the microstructures when viewed at different magnification levels: The high aspect ratio carbon posts add complexity to a flat surface on the microscale while the different carbon nanotubes and nanofibers add a nanoroughness to the otherwise flat surface of the carbon electrodes.
22
Plane
C-MEMS
C-MEMS Post Arrays
CNTs/C-MEMS
Figure 1-9:. Fractal geometry of Carbon MEMS electrodes as part of a scheme to increase surface area
This concept could be even pushed further with the application of engineering that specifically increases yet even more surface area on the nanofibers (e.g. soft plasma etching treatment). The fractal concept is schematically represented in Figure 1-9. Carbon MEMS (C-MEMS) has been in current active development within the Madou group for several years. There is need for work on the fundamental issues of the properties of the Carbon MEMS material before going further in the possible application of that technology. The aim of this work is to further push the understanding of nano/micro Carbon MEMS by identifying the different fabrication methods that can lead to such structures, studying their in depth properties (physical, chemical, electrochemical) and understanding what those properties mean for the different applications where these microsystems can be useful (chemical storage, implantable electrodes, growth medium for neural growth, glucose detectors, etc.)
23
250 x
3k x
30k x
300k x
Figure 1-10: different orders of a Koch curve seen in parallel with carbon fractal electrodes seen at different magnifications. For illustration purposes, multiple orders of the Koch fractal curve is shown in parallel with the multiple orders of our fractal carbon electrodes. At every step in the generation, the Koch curve gains complexity by replacing any straight line with a line and triangle. The new shape has obviously more surface area than the shape in the previous iteration. Our fractal carbon electrode mimics that same mathematical process. The electrode posts add complexity to the flat carbon electrode. The carbon nanotubes add complexity to the electrode posts. The bamboo like carbon defects in mutli walled carbon nanotubes add complexity to the carbon nanotubes’ surface. Note that when observed at different scales, the fractal carbon electrode looks almost the same (compare 250 and 30k magnification). It is said to be self similar, which a common attribute of fractals that are present in nature.
The usefulness of such a fractal construction, if one is capable of perfect control over the fabrication of such a structure, has been identified by a first order analysis performed on an electrochemical system and the details have been described in a manuscript [19]. The 24
geometry considered was that of a space filling branching fractal shown in Fig1-11. The main result of the study concluded that a similarly constructed fractal electrode would have a scaling more akin to sheets than to volumes. In more practical terms, the total resistance of a volumetric system sees it’s resistance drop to 80% of its value when the electrode volume of is doubled, while for an electrode constructed in a fractal architecture, the total resistance would drop to 50%. That difference shows illustrates how fractal architecture can be an advantageous tool for scaling systems in an efficient manner. The total resistance is actually seen to grow smaller as a function of the number of the smallest elements in the tree.
Figure 1-11: graphical representation of a branching fractal showing three fractal levels. That construction was used in [19] to validate the efficient scaling of Fractal electrodes.
The main results from the model are shown in Table 1-1. 25
Table 1-1: The main results from the zeroth order modeling of a fractal electrode
The analysis suggests that a space filling fractal like electrode can offer regular electrodes by minimizing the internal resistance of an electromechanical system being considered.
26
Chapter 2
2.Physically Modified C-MEMS Processes
2.1.
Rapid
Freestanding
SU-8
Photolithography
Process and its Application in C MEMS
This chapter describes a simple novel method for fabricating freestanding SU-8 microstructures. Unlike conventional methods for obtaining freestanding microstructures that rely on sacrificial layers to perform the release from the substrate, the method does not involve the use of a sacrificial layer but is based on a sacrificial substrate made out of polystyrene (PS). The release method is purely mechanical and can therefore achieve in a single minute what would normally take hours to do using chemical etching techniques.
27
2.1.1. Introduction
One of the major enabler of surface micromachined MEMS devices has been the use of sacrificial layers to liberate mechanical structures from the underlying substrate. Small cantilever devices, micromirrors, resonators, gyroscopes, and accelerometers are being fabricated using surface micromachining techniques that rely on sacrificial layers. The most common sacrificial materials used are non organic materials like silicon dioxide (SiO2) and phosphosilicate glass (PSG). Hydrofluoric acid (HF) in an aqueous solution is commonly used to selectively etch PSG and SiO2 in the presence of silicon or silicon nitride. PSG can thus be etched using aqueous HF in pattern sizes of up to 2000 μm [20]. Commercially available MEMS prototyping processes such as PolyMUMPs by MEMSCAP© , NC require etch holes every 30 μm in order to effectively etch away PSG deposited using low pressure chemical vapor deposition (LPCVD), and free the polysilicon microstructures [21]. Aqueous HF is a toxic and hazardous material and is not safe to be handled by the inexperienced user. Aqueous HF can also easily attack metals such as aluminum or titanium and these can therefore be used as sacrificial materials too. Non organic sacrificial materials are less common. Photoresists widely used in the research environment, such as Shipley® 1800 series, can act as sacrificial materials in methods such as lift-off techniques for patterning metals [22]. Acetone is used in that case to remove the photoresist from underneath a metal layer. Similarly, another Shipley photoresist SPR220.7.0 has been used as a sacrificial material for SU-8 structures [23]. Propylene glycol methyl ether acetate (PGMEA) which is 28
normally used to develop SU-8 was used as the etching agent. Free standing SU-8 microfluidic devices have been previously fabricated by bonding multiple layer of SU-8 patterned on silicon wafers with thermally grown SiO2 [24]. The detachment of SU-8 was achieved by using HF (50%) as an etching agent at rate varying between 5 and 30 hours for a 4 in wafer. Whitesides et al. have recently used inorganic water soluble polymers as sacrificial materials for the fabrication of freestanding SU-8 structures [25]. Both dextran (a branched polysaccharide) and poly(acrylic acid) (PAA) were spin coated unto a HCl treated silicon wafer. Water was then used to dissolve the layers between SU8 and a silicon wafer to create overhanging structures. The process used would normally require at least 2 hours to detach a layer of SU-8 from a 4 in wafer substrate. The method presented in this work, Rapid Freestanding SU-8 (RFS), allows the release of 4 in layer of SU-8 from the underlying substrate in almost one minute time. By simply peeling the layer of SU-8 from the surface of a polystyrene (PS) Petri dish substrate, free standing micropatterned films of SU-8 ranging between 10 μm and 2mm can be obtained. The weak adhesion between PS and SU-8 allows Petri dishes to be used as an inexpensive temporary substrate for free standing SU-8 films. The technique can be used to simplify the fabrication of microfluidics, drug delivery patches [26], microfilters etc. All the advantages of High Aspect Ratio SU-8 structures can be transposed onto flexible SU-8 films. The ease and simplicity of the technique makes it amenable to mass manufacturing.
29
2.1.2. Materials and Methods Disposable Polystyrene Petri dishes (size 100 mm × 15 mm, with vertical stacking rings) were purchased from Sigma Aldrich, MO. TECHNI TOOL™ Stainless Steel wafer tweezers were purchase from Fischer Scientific™, PA. NANO SU-8 of different compositions was purchased from MicroChem©, MA. PGMEA (>99.5%) used as SU-8 developer was purchased from Sigma Aldrich, MO.
Photolithographic Process The standard practice for photolithography using SU-8 as widely reported in the literature is used. The only significant difference is that the bake steps are performed at 100oC instead of the usual 95oC to account for the temperature gradient that the separation between the surface of the Petri dish and the hot plate imposes. The process is described for the example of a 15 μm SU-8 layer. SU-8 25 is first spin coated onto the surface of the PS Petri dish (the top cover). The spin profile consists of 5 seconds at 500RPM followed by 30 seconds at 2000RPM. The ramp up acceleration does not to exceed 350RPM/sec. No surface treatment is necessary to ensure the proper adhesion of the SU8 on the PS. The sample is then placed a hot plate. A single step soft bake at 100oC for 15 minutes is performed. The sample is then exposed to a 150mJ/cm2 dose (25 seconds at 6mW/cm2) of near UV radiation from a mercury lamp through a patterned photomask. A 4 minutes post exposure bake on a hot plate at 100oC follows. The sample is left to cool for 5 minutes then the 15 μm SU-8 layer is removed from the top of the Petri dish and developed in a PGMEA bath. 30
SU-8 layer removal process The side of the Petri dish is first broken with scissors. A little vertical cut into the top PS layer is then performed. This normally creates a little space in between the SU-8 layer and the PS substrate where the flat tip of wafer tweezers can be inserted. The tweezers can be slid in a full circle motion along the edge of the Petri dish so as to liberate the edge of the SU-8 layer (see fig.2-1), after which the layer can be easily peeled. The removal process normally can easily be performed in less than a minute. More than 90% of the top surface of the Petri can be recovered in a single piece after peeling process. The sample can then be developed in a PGMEA solution.
Figure 2-1: Photograph of the SU-8 layer removal process showing the full circle motion required to release the edge of a 400 μm layer. (A thick darkened layer with ripples is shown for visualization purposes, since the usual SU-8 sheet are transparent and don’t show well in photographs)
Pyrolysis process
31
A standard C-MEMS process is used as described in details in [5] [27]. In brief, micropatterned sheets of SU-8 are pyrolysed in a quartz flow through furnace under N2 atmosphere. The SU-8 samples are place in the furnace sandwiched between a quartz plate and a steel weight (1/8 in. shim stock plate of almost the same size) to keep them from deforming during the pyrolysis process. A hard bake is first performed at 200oC for 30 minutes. A ramp up to 900oC at an average rate of 10oC/min follows. The sample remains at 900oC for an hour and then is left to cool down to room temperature (approximately 8 hours). The process converts SU-8 into amorphous carbon that can be used in electrochemical applications.
2.1.3. Results and Discussion The poor adhesion of SU-8 to Polystyrene allows for the easy peeling of the SU-8 layer from the top of the Petri dish. The low tensile strength (50MPa) of PS allows it to be easily broken with scissors and expose a separation between SU-8 and the top surface where tweezers can easily be introduced and the SU-8 layer peeled. PS is yet stiff enough (young’s modulus close to 3.5GPa [28]) to be used as a substrate and mechanical support for the SU-8 layers during the spin coating and exposure steps. Any other polymer with the same mechanical characteristics as PS can be used in the same process as long as it has low adhesion to SU-8. Polyimide is a candidate since it has been reported to have bad adhesion to SU-8[29]. The commercially available sheets of 125 μm are too flexible to act as a substrate and are over a hundred times more expensive than PS. Teflon has been previously used in a similar application but required a very specific plasma treatment in order to balance out the adhesion of SU-8 on its surface [30]. Polystyrene substrate is 32
unique in the fact that not preprocessing or postprocessing is required, and the substrate can be used as is (even reuse of the disposable Petri dishes has been attempted successfully).
SU-8 Sheets Micropattern SU-8 sheets using the described fabrication technique were fabricated from 10 μm to 400 μm. The thinner films are quite flexible and easy to handle but can tear easily. The thicker films are more rigid and quite resilient. Longer soft bakes are recommended for thicker layers. Adverse effects can occur when a thick layer is not soft baked long enough. The top layer can crosslink much easier in that case and cause a rippled surface structure such as the one seen in Fig.2-1. This problem does not arise at all when the softbake is done for the right amount of time (1 hour 45 minutes for a 400 μm SU-8 layer). SU-8 sheet can be easily cut or stamped to any desired shape just like regular dry film photoresist. They can also be bent to radii of curvature smaller than 2 cm without fracturing or incurring plastic deformation (see Figure 2-2).
33
Figure 2-2: Photograph of a cut strip of SU-8 layer bent to a radius of curvature of 2cm. The layer shown was 20 μm thick The SU-8 lithographic process can easily be repeated multiple times with one layer serving as a base for high aspect ratio structures. Reference [26] describes how to fabricate arrays of hollow needles such as the one used in drug delivery patches. The process uses a combination of PDMS and SU-8 process which can be greatly simplified using the RFS technique. Since the peeling step occurs before the small SU-8 features are developed it is possible to render even the smallest suspended structures without having to worry about either stiction to a substrate or long release times typically associated with conventional release processes. Multiple freestanding layers of SU-8 can also be bonded together to form microfluidic devices similar to those described in ref [24]. It was observed that under the effect of time (circa 40 days), SU-8 sheets turn from a completely transparent to brownish and loss their flexibility dramatically. It is not quite clear what causes the darkening in the SU-8 but it is speculated that crosslinking of SU-8 34
continues to happen in ambient conditions and leads to increased brittleness in the material.
Sample fabricated devices Simple structures as a proof of concept of the method were first tried using a two photolithographic step process. The first layer was a 30 μm SU-8 and was completely exposed to UV. The second layer was 100 μm thick and was exposed through a photomask. Some resulting structures are shown in Fig. 2-3 and Fig.2-4.
a)
b)
Figure 2-3: Example of free standing SU-8 microstructures. a) SEM image of 100 μm SU-8 posts on flexible 30 μm SU-8 substrate fabricated using RFS. b) photograph showing the same array of posts (middle square with a 1cm side) as it can be used for patch drug delivery applications
35
a)
b)
Figure 2-4: Another example of microstructure obtained using RFS. a) SEM image of an circular array of SU-8 microposts on a flexible 30 μm SU-8 substrate. b) Photograph of the same array attached in a flexed position on a Si/SiO2 wafer.
While it’s easy to imagine a large number of applications that can use RFS, it was originally conceived while trying to fabricate freestanding Carbon MEMS structures for use as anodes for lithium ion microbatteries. The added surface area from the microstructuring of carbon films can increase Li ion intercalation rates in and out of the carbon film. This will be discussed later in this dissertation (see Section 5.1). Figure 2-5 shows an example of such a structure. RFS was performed using a single lithography step on a 75 μm thick SU-8 layer. The carbon layer is almost transparent where the holes are located. The SU-8 sample was inadvertently not completely developed which was not very obvious when the sample was still made out of SU-8 since the film is transparent in that state.
36
a)
b)
Figure 2-5: C-MEMS thin sheets of carbon for Li ion anode applications fabricated by pyrolyzing sheets fabricated using RFS. a) photograph of the thin sheet of carbon with an array of holes made using a one step RFS followed by Pyrolysis. b) SEM image of the same sample showing the 30 μm holes in the thin film of carbon (note the sample was under developed and some holes are not completely empty).
2.1.4. Conclusion We have introduced RFS as a novel technique for obtaining freestanding SU-8 microstructures. This technique can be applied in a variety of research and commercial applications. The simplicity of this technique makes it highly probable that it be well adopted specifically by the microfluidics community.
37
2.2.
A
Stretching
Flow
Polymer
Deposition
Technique for the Fabrication of Suspended Carbon Nanowires There has been great interest in nanoscale carbonaceous material since the discovery of Iijima [11] of carbon nanotubes. Fascinating material properties have been attributed to these nanofibers from superior mechanical, chemical and electrical properties. Studying and isolating carbon nanofibers has so far been a complex process which involved very low process yields. The carbon fiber had to be grown first on a surface and then further steps had to be taken to isolate the given fiber and contact electrically or submit it to mechanical tests. Making full use of these enhanced properties of carbon nanofibers is contingent upon finding fabrication processes that can control the growth of single fibers in specific places. Effort along these lines have been pursued [31], [32], [33], but with little success. Unless newer techniques are found the commercial success of carbon nanofibers is compromised by the very expensive and erratic process used to obtain in a repeatable manner. Richard Feynman, the father of microfabrication stipulated the possibility of fabricating machines that can fabricate smaller machines. In this work, a solution based deposition process is a process in which a combination of small machines (microposts) are used to fabricated smaller machines (nanowires ) using the combination of flow and surface tension.
38
2.2.1.
Introduction
Suspended nanowires and microwires formed with a SU-8 photoresist have been reported earlier in the literature [5, 34] . While in [34] these structures were created using a combination of photolithography and e-beam lithography, no convincing explanation was ever given by [5]. This work recreates the conditions in which suspended nanowires can be placed and removed in a repeatable manner on a pre-fabricated array of SU-8 microposts. The deposition process is based on the stretching flow of a PGMEA solution supersaturated with SU-8 resin in between two SU-8 microposts.
2.2.2.
Materials and Methods
SU-8 Microposts Array Fabrication The fabrication of the micro array of SU-8 posts was done in the conventional SU-8 photolithography manner used for SU-8. The parameters for the fabrication are shown in table 2-1. The two different types of SU-8 were purchased from MicroChem©, MA.
Table 2-1 Parameters for SU-8 microposts array fabrication
Nominal
Type
of Pre-spin
Thickness SU-8
Exposure Time Post Exposure
Softbake
RPM/
Time
Spin RPM
95oC
at Using 6mw/cm2 Time at 95oC UV Source
5 μm
SU-8 5
500/3000
5 min.
20 sec.
2 min.
150 μm
SU-8 100
500/2000
90 min.
120 sec.
16 min.
39
The typical SU-8 process consists of: Spin coating the SU-8 on the surface of a wafer to obtain uniform thickness, soft baking the sample to allow the solvents to evaporate, Exposure through a mask to activate the photoacid generator, post exposure baking to drive the crosslinking of the SU-8 epoxy layer. In our case, a 5 μm layer of SU-8 was first prepared on a Si/SiO2 4 in. substrate according to the parameters shown in table 2-1. The 5 μm is used as base layer for a good adhesion of the thicker 150 μm that was used to form the microposts. The base layer is flood exposed and post exposure baked before the application of the 150 μm layer of SU-8. Having a base layer ensures that the microposts don’t lose adhesion with the substrate under the repeated immersion in multiple solutions used in the fabrication process. The second layer as fabricated thickness is 165 μm (seen from SEM pictures).
Supersaturated Solution of PGMEA and SU-8 Resin
A 300 μm layer of SU-8 was spin coated onto a silicon wafer. The wafer was soft baked for 2 hours to allow the solvent (Gamma Butyrolactone GBL) to evaporate from the layer. The wafer is then placed in a beaker where 15 ml of Propylene Glycol Methyl Ether Acetate (PGMEA, 99% Sigma Aldrich), the typical developer solution of SU-8. Mild agitation is used to allow the PGMEA to dissolve the uncrosslinked layer of SU-8. 40
The resulting mixture is the poured in a dropper container for use at a later time in the nanowires forming experiment. The resulting solution of Supersaturated PGMEA SU-8 is referred to as SP SU-8
Fabrication of SU-8 Nanowires
The wafer sample containing the microposts array was placed in a shallow beaker and secured under a microscope. A droplet of SP SU-8 is placed adjacent to the microposts array. Isopropanol (from Sigma Aldrich)is then flowed from behind the droplet of SP SU8 towards the microposts array using a high density polyethylene squirt bottle dispenser (Mc Master Carr) providing an approximate free flow velocity of 10 cm/s . The Isopropanol flow entrains droplets of SP SU-8 to the top of SU-8 microposts where they are partially absorbed to the surface. Further flow of Isopropanol stretches the small droplets of SP-SU-8 into filaments which stretch from one micropost to the other. SP-SU8 entrained by Isopropanol appears to form spherical, sheet like, and filament shapes. The stretching of SP-SU-8 by the hydrodynamic shear and momentum forces exerted on it by the Isopropanol creates the typical necking profile that is observed in conventional stretching flow experiments. Longer exposure of thin layers SP-SU-8 to Isopropanol (longer than 10 sec.) allows the exchange of Isopropanol and PGMEA by diffusion, convection and surface tension forces. Isopropanol has a great wetting ability for SU-8 (contact angle< 20 degrees ) [35]. The solubility of SU-8 in Isopropanol is limited. When PGMEA is displaced, the SU-8 resin precipitates into a quasi solid phase that is strong 41
enough to sustain further Isopropanol flow as well as blow drying using a nitrogen gun. The nanowires thus formed can either remain there permanently or be removed again using immersion in a PGMEA solution. Different compositions of SU-8 mixtures can make it more or less easy to form these nanowires. The calculated molecular weight of SU-8 resin monomer (C87H96O16)is 1396 g/mol[36], but a distribution of molecular weight exists in the actual commercially available products with a molecular of approximately 6060 (be it the Epirez SU-8 from Shell company[37], Hexion, or the SU-8 supplied from MicroChem). It is suspected that the larger molecular weight mixture that occur when a developer is used repeatedly are more favorable to the formation of SU-8 nanowires since they allow for longer range cohesion and effectively a higher viscosity.
Pyrolysis
The SU-8 nanowires, fabricated using the stretching flow technique, are made of uncrosslinked SU-8 resin. These nanostructures can be carbonized by heating to around 900oC for 1 hour under a non reactive N2 atmosphere, in a process detailed in [38]. The resulting carbon structures are 80μm long and can be as small as 200 nm in diameter.
42
2.2.3.
Results and Discussion
Stretching Flow A typical stretching flow occurs when a liquid is put in between two disks and the disks are separated at a constant velocity from each other[39],. The disks can be initially submerged in a solution that has the same density as the liquid being stretched so as to offset gravitational effects. This method as well as elongational flow has been used to characterize the viscoelastic properties of polymers and polymer solvent mixtures [40]. The variables in such an analysis include the dimensionless parameters. Stretching flow normally leads to necking of the fluid at distances smaller than 10 cm of elongation. The necking is often noted but no clear data is available on how small the neck of the wire gets right before breaking. Inertial and frictional forces are countered by surface tension forces until the latter become too weak to sustain increasingly shrinking filament of liquid. Breaking occurs at that point. In our case the stretching occurs between a micropost and another driven by kinetic forces (shear and momentum) of the other liquid phase (Isopropanol) on the surface of the SP-SU-8 mixture. While the smaller filament radius normally leads to necking and then breaking, in our case, smaller filamentous radius means that the PGMEA that is dissolving the SU-8 Resin inside the filaments can be easily exchanged with Isopropanol, because of the enhanced surface to volume ratio and lead to a precipitation of SU-8 resin observed through a hardening of the filaments in the flow. These effects occur at different time scales but they allow enough flexibility window to repeatedly obtain SU-8 nanowires suspended in between microposts that are 43
100 μm apart. The nanowires can be completely removed by immersion in a PGMEA solution. In our case what is different is that the stretching occurs with one fixed end and the other end being stretched away by hydrodynamic forces. Once the floating wire arrives to the second post anchor, it becomes more stable and less prone to the hydrodynamic forces. Surface tension in this case takes over the stretching force and brings the filament to its lowest energy form. The same surface tension effects make it more likely for a micropost with a T-topped profile to hang the nanowires at its top. If the separation between posts is very small, it is sometimes very hard to properly rinse the SP-SU-8 in the presence of high aspect ratio structures. It is believed that SP-SU-8 occurs in a regular development process in two instances. First, when the developer PGMEA solution is repeatedly used from one run to the other (which we believe is what caused the presence of these nanowires as artifacts). Second, when very high aspect ratio SU-8 microstructures are being developed and there are very hard to reach locations for Isopropanol to rinse away. As seen in Figure (see Fig.2-6).
44
Figure 2-6: SEM image of very high aspect ratio structures showing the formation of ridges from by PGMEA saturated with SU-8 resin suspended in between posts. The isopropanol rinse could not in this case, reach the bottom of the high aspect ratio crevasse and rinse away the mix. If the stretching is fast enough, a constant diameter area occurs in the case of mechanical stretching seen in Figure 2-7.
45
Figure 2-7: Showing the stretching flow profile for different stretching rates for a viscoelastic poly acrylamide 0.9% in water. 1 mm/s on the left and 8 mm/s on the right. Observe the plateau in the middle for the faster rate. [39]
46
(a)
(b)
(c)
(d)
Figure 2-8: Optical microscopy capture sequence of the formation of the nanowires. (a) a microposts array is shown in a PGMEA shallow bath with a droplet of SP-SU-8 deposited on the right side. (b) Isopropanol is flowed in the general direction from right to left using a squirt bottle, the squirt time last less than three seconds. (c) and (d) The darker areas show where the bridges between posts have occurred. In some places the SPSU-8 remains at the bottom of the posts while in others it’s seen to occupy the top area. This is caused by the not very controlled free flow nature of the Isopropanol flow. The center to center distance between microposts is 100 μm
47
(a)
(b)
(c)
(d)
Figure 2-9: The stretch of the filaments is seen to correspond to the streamlines dictated by the streamlines within and outside the array. The center to center distance between microposts is 100 μm
The leading edge of the filaments is not necessarily a thin wire, it is more likely to be a larger mass of SP-SU-8 which is more easily driven by the hydrodynamic forces of the Isopropanol phase. These larger masses (spherical or sheet like) can cover more than one post a the same time and attach to the top of the posts in the manner shown in Figure (see Fig. 2-10). These larger masses can either be thinned down by the hydrodynamic forces if 48
the flow is strong enough before the mass becomes stronger from the displacement of PGMEA by the Isopropanol (which take around three seconds in this case. Further flow of Isopropanol after that 10 seconds approximate period leads to no change in the shape of the filaments which become quite rigid after precipitating into the Isopropanol solution. Epon SU-8 resin, the main resin forming SU-8 photoresist, is known to have a melting temperature around 80oC, [41]. Assuming the Isopropanol is capable of displacing all the PGMEA, the solid obtain is the SU-8 resin alone.
Scanning Electron Microscopy of SU-8 Nanowires
Figure 2-10: SEM image of an SU-8 nanowire stretched between two microposts. The Isopropanol flow direction was coming for the top of the picture downwards. The asymmetric shape can bee seen testifying to the original attachment of SP-SU-8 on the upstream micropost 49
Figure 2-11: SEM image of an array of SU-8 nanowires bridging the tops of SU-8 microposts. The majority of the wires are in the general direction of the flow (top of the Figure downwards). The crossflow wires can occur when larger masses of SP-SU-8 land on multiple posts at the same time and local divert the flow stream lines. In the presence of slower Isopropanol stretching flow, surface tension can take over easily and form sheets of thin wires depending on how many posts the SP-SU-8 covers.
50
(a)
\ (b) Figure 2-12: larger masses of SU-8 seen spanning multiple microposts in a sheet format. These masses are possibly more solid than others by the time they got attached to the array (because of longer exposure to Isopropanol) and therefore not be influenced by further hydrodynamic forces. This fact explains their coexistence with smaller wires. (b) The wires are shown not to always be at the very top of the post, by they always occur between the closest two points in on the posts (because of surface tension forces’ drive to minimize the area). This is why T-topped structures are a very good choice to amplify that effect.
51
Scanning Electron Microscopy of Carbon Nanowires
(a)
(b) Figure 2-13: SEM image of the pyrolyzed SU-8 nanowires. (a) The nanowires spans a distance of almost 100μm and the minimal diameter observed on it is 250 nm. (a) shows the example of the carbon nanowire 250 nm thick. 52
Figure 2-14: SEM image of an array of carbon nanowires bridging C-MEMS posts. The nanowires are strong enough to bend the posts as the carbonization progresses. The bending is approximately 15o. The top view of the posts show the top of a shown the leaning profile of the various posts. The deflection probably occurs below 500oC before the SU-8 is transformed into carbon, at a point where the structure is still flexible enough to deflect without breaking.
Pyrolysis SU-8 Bisphenol resins can be crosslinked by mere heating. While the melting temperature of Epon SU-8 is 80o, it appears that the flow at this temperature is not enough to destroy the preformed structure formed in the stretching flow method. In fact the nanowires crosslink quickly enough to bend the SU-8 microposts to which they are attached (see Fig.2-14). Preliminary data on these nanowires from electrical probing have
53
shown a 15K ohm resistance as well a possibly strong piezoresistive effect. These effects require further validation and characterization.
Method Discussion and Proposed Controlled Formation Method
The proposed stretching flow can be tweaked to yield more uniform results by adjusting the following parameters: - Resin concentration of the SP-SU-8 - Replacing Isopropanol by a mixture of Isopropanol and PGMEA (1:10) allows a larger time for the nanowires to form, prior to the precipitation of the resin. - Doing the stretching experiment in a microfluidic channel.
The stretching flow that generates the nanowires in a bulk solution can be miniaturized to yield a much more directed fabrication method. The proposed technique involves using a coaxial flow of isopropanol and SP-SU-8 that emerges from the tip of a pipette to form the desired elongational flow. The pipette can be attached to a micromanipulator that can be used to dictate the exact location of the desired microspots
2.2.4. Conclusion The mechanism of the formation of suspended SU-8 nanowires in between SU-8 microposts has been explained and reported in an experimentally repeatable fashion. The 54
nanowires are generated by the stretching flow of a PGMEA solution saturated with SU-8 resin driven by the hydrodynamic forces of the rinsing Isopropanol flow. These nanowires can be converted into carbon nanowires by simple pyrolysis in a non reactive environment. The resulting assembly is a 250nm suspended carbon device that can be used as a sensing electrode for various electrochemical applications. One of which is attachment modulated field effect transistor or resistor that is completely separated from surface effects normally present because of the fabrication on substrate.
55
Chapter 3
3.Electrochemical Modifications to CMEMS Techniques The importance of fractal structures is that they provide a physical context that allows access to the power of nanomanufactured materials. Nature is capable of fabricating fractal like structures in the gaseous, solid and gas phase. The same eddy currents that help you mix the sugar in your tea, drives the large swirling piles of autumn leaf outside you house, forms small tornados, and generates large hurricanes in tropical areas. While nature has its own driving force for each of these phenomena, the key in imitating such self similarity at such a wide scale is controllability. The level of entropy increases when going from solid to liquid then to gas, and with it decreases the controllability of the system. The aim of our nanomanufacturing effort is not just create fractal like solid 56
structures, but to do so in an addressable fashion. Doing so in the solid phase provides the most control but the mechanical tools available for these techniques are too big (mechanical milling), or too slow (AFM tips). Solid phase machining is in a sense too rigid to provide a viable solution. The higher degree of flexibility is provided by the liquid phase. Since our desired end-structure is a solid, a good compromise between flexibility and controllability is to use fabrication techniques that are at the interface between the liquid and solid phases. Machining Techniques at the interface of solid and gas (or plasma) such as sublimation, physical and chemical vapor deposition, and reactive ion etching, are useful but often too slow, result in two dimensional structures, and are hard to direct in 3D. The addressability constraint makes it such that the optimal machining technique has to be triggered by the solid substrate in a selective manner. Examples of fabrication techniques that satisfy these requirements are: -
Surface specific precipitation (such as electroless deposition or crystallization)
-
Electro kinetic assembly methods, such as aggregation through electrophoresis used in electrophoretic painting or dielectrophoretic particle trapping.
-
Self assembly on hydrophilic/hydrophobic patterned surfaces.
-
Thermally driven polymerization on thermal conductor/thermal insulator patterned surfaces.
-
Electrodeposition techniques
Another important requirement for choosing an efficient technique to fabricated fractal like structures is that the method should be easily expandable to 3D.
57
From all the described machining and deposition methods electrochemical based deposition and etching is a tool we favor for the following reason: -
High speed of deposition and etching (easily microns per minute)
-
Amenable to 3D structure fabrication
-
Easily addressable through the use of electronically conductive electrodes on insulating substrates
An example of electrochemical deposition is electropolymerization where organic monomers are deposited onto an electrode from a solution. We will use the example of The electropolymerization of pyrrole on C-MEMS structures as an example to illustrate the versatility of this method in fabricating fractal-like structures.
3.1.
PPY C-MEMS
3.1.1. Introduction Polypyrrole is part of a family of special conductive polymers that have very interesting applications in displays[42], drug delivery[43], tissue engineering[44], etc. Pyrrole monomer (C4H5N), seen in Fig. 3-1, Can polymerize at the surface of an electrode through an oxidation process in an aqueous solution in the presence of a surfactant molecule, often Dodecylbenzene sulfonate DBS. The linear chain that results attaches well to metallic and carbon surfaces.
58
N
Figure 3-1: Pyrrole monomer C4H5N PPY C-MEMS is a combination of microfabrication technique that results in fractal like structures. C-MEMS microstructures are first fabricated through the regular C-MEMS previously described to make 85 μm and 45 μm high posts with a 25 μm diameter. This carbon structure is then used as a substrate for deposition of PPY after which the sample is put again through pyrolysis in a non reactive atmosphere. The electrodeposited PPY in it’s various states is thus carbonized. PPY is a good choice of materials for this process for several reasons. First it is a conductive polymer that can be deposited ad infinitum. Although the polymer is a linear chain, it has a strong interaction between neighboring ring structures, which makes it to char before melting, therefore maintaining its general shape during the pyrolysis process. Features as small as 20 nm can be conserved given the right processing conditions.
59
3.1.2. High Overpotential Electrodeposition of PPY Standard conditions for the electrodeposition of a good quality film of PPY consist of using an electrodeposition solution of 0.1M NaDBS and 0.1M PY (pyrrole monomer). Electrodeposition is normally performed at 0.65V Vs Ag/AgCl reference electrode. This results in a current density close to 0.1mA/cm2 and a deposition rate of 1 μm/h. In our case, a similar process was used but a deposition voltage of 5V was used. Electrodeposition at this high an overpotential results in both increased effect of electromigration as well as pronounced effect of hydrolysis. At these potentials, oxygen is being produced at the anode (our deposition electrode in this case) and hydrogen is evolving at the cathode (large gold counter electrode in this case). The 5V electrodeposition resulted in a 35 mA current through the 2 cm2 electrode. Electrodeposition was sustained for a period of 1 minute at the end of which the current increased to 40mA. The resulting structures are shown in Fig.3-2
60
(a)
(b)
(c) Figure 3-2: PPy deposited on top of C-MEMS posts using high overpotential deposition at 5V vs. Ag/AgCl. The bare C-MEMS posts area shown in (a). The PPy coated posts are shown in (b) close up of the mushroom like smaller posts covered with a 5-10 μm PPy Layer. The effect of increased field intensity at the top of the posts is reflected by the higher deposition rates. An occasional bubble seems to be caused by a local accumulation of O2 gas that did not get dissipated in the solution. Using similar testing conditions but 61
changing the concentration of PPy by ten fold up to 1M yielded some of the structures seen in Fig. 3-3. The higher concentration of Py monomer dampened out the mushroom
(a)
(b)
(c) Figure 3-3: PPy covered C-MEMS structures using high overpotential deposition (5V), as well as higher concentrations of Py monomer (1M). The PPy covered microposts are shown in (a). The cauliflower like deposition at the edge of the film is shown in (b). In (c), a close up of the cauliflower surface shows a similar granularity at the 200nm level. A fractal like structure indeed.
62
effect that was observed in the 0.1M solution, possibly because of the shielding effect that Pyrrole as a relatively polar molecule has on the electric field [45]. These high concentration monomer deposition conditions resulted in larger blobs or sphere like deposition which at the edges of the carbon film looked a lot like 200 μm cauliflower. The close-up view shows that at the small scale the deposition film exhibited smaller spheres of 200 nm approximate dimensions. Being able to control this deposition requires a full control of the intensity of the field inside the electrodeposition solution. Nevertheless we were able to obtain a self similar structure with feature spanning three orders of magnitude in a single deposition process (from 200 μm to 200 nm).
3.1.3. Templated Electrodeposition of PPy
Hydrogen Bubble Templated Electropolymerization When the high overpotential method was applied for the electrodeposition of PPy, H2 bubble were generated at the counter electrode and migrated towards the carbon surface on which PPy was being deposited. A high deposition rate was observed at the surface of these bubbles and inverted dome shaped structures were also obtained. are shown in Fig. 3-4 and 3-5.
63
(a)
(b)
Figure 3-4: PPy deposition on the surface of hydrogen bubbles adsorbed on the surface of carbon. 100 μm PPy half bubbles are shown in (a). 5-10 μm smaller PPy bubbles
Figure 3-5: Spherical like shapes of electropolymerized PPy using hydrogen bubbles as a template. Figures 3-4 and 3-5 indicate that these structures are hollow from the inside. Similar results
have
been
reported
with
different
electrodeposition
solutions[46].
Electropolymerization of PPy seems to occur faster on specific surfaces than in the solution. It is speculated that the probability of interaction of adsorbed with oxidized 64
monomer of pyrrole is much higher on the surface because of the enhanced surface diffusion rates compared to the diffusion rates in the solution. This seems to be true especially for the case of Polytetrafluoroethylene (PTFE).
PTFE Templated Electropolymerization. PTFE or Teflon when extruded into a thin tape format, present a nice polymeric crystal shape with long parallel strands perpendicular to the surface of the tape. The structure can be seen in Fig. 3-6.
(a) (b) Figure 3-6: SEM image of the surface of a PTFE tape showing bunches 200 nm diameter fibers terminated at the top by larger 2μm spheres (b). The bunches seem to be separated by 10-15 micron gaps (a). The top spheres holding the vertical strands together don’t seem attached, which under in plane stress can elongate dramatically and align themselves with the direction of the tape. A similar sheet of PTFE tape was slightly stretched and used to cover the surface of a CMEMS carbon sheet. The sample was placed in standard PPy electrodeposition solution/ The tape/carbon assembly was lose enough to allow the infiltration of electrolyte which 65
both surface very well. A 5V potentiostatic deposition for 1 minute was used to create the structures seen in Fig. 3-7. A small forest of PPy formed from underneath the Teflon tape. It is not well understood how the branching occurs at very defined positions on the PPy structure, but it is highly likely that not quite conformal to the shape of the teflon but also the shape of the electric field profile inside the teflon tape. The end result would be a combination of diffusion effects as well as electric field shaping.
(a)
(b)
(c)
(d)
Figure 3-7: SEM images of the same area of a small PPy forest electrodeposited in a template made of PTFE tape. The photo shows the self similarity at different scales. The magnification is 300, 1.3K, 4K, 13K for (a), (b), (c), (d) respectively.
66
(a)
(b)
Figure 3-8: Electropolymerization of PPy using a stretched Teflon tape. The tape was stretched in the horizontal direction. In fact where is apparently more stretched, the growth of these structures seems directed by the direction of the extension of the strands seen in Fig. 3-8.Intricate structures with very interesting complexity are seen to occur, similar to the ones in Fig 3-8(a). work is ongoing to be able to fabricate space filling structures that present this self similarity related to truly fractal Structures. Most importantly the feature shown in all these figures can be transformed into glassy carbon which is anywhere between 100 to 10000 times more electronically conductive than PPy (typically between 40 and 0.01S/m) [47, 48].
67
(a)
(b)
(c)
(d)
Figure 3-9: Pyrolyzed Electrodeposited Polypyrrole. The resulting carbon structure displays 3D self similarity typical naturally occurring fractals. The carbon resulting from the pyrolysis of PPy shows structural shrinkage and some loss of mass due to etching when compared to the initial PPy structures. Tip sharpening is an indicator of a kind of corrosive action occurring through pyrolysis. The shown tip is as small as 20nm and can possibly have proximal probe microscopy applications. The etching effect can be tailored by using different pyrolysis conditions and more or less be eliminated.
68
3.1.4. Conclusion Electropolymerization of Pyrrole can be used in conjunction with C-MEMS to obtain PPy and carbon self-similar structures spanning more that 3 orders of magnitude of dimensional difference. These techniques are tools towards the fabrication of Multiscale C-MEMS that can leverage all the advantages of fractal structures from the nanoscale to the macroscale.
3.2.
Tools for Fractal Micro-Electrode Fabrication
Using a Combination of Photolithography and Electrodeposition.
3.2.1. Introduction
The electropolymerization of Pyrrole (Py) is a process normally used to deposit thin films of electrically and ionically conducting polypyrrole (PPY) for various applications. Solutions of surfactants are commonly used to ensure the deposition of a flat and smooth surface in a timely manner. Anionic surfactants, such as sodium dodecylbenzenesulfonate NaDBS [49] and sodium dodeyl surfate (SDS), are commonly are used for thin film fabrication of microactuators. Cationic surfactants such as Cetyl trimethylammonium bromide (CTAB), have been shown to produce ribbon and sphere like PPy nanostructures 69
in chemical oxidation driven polymerization using ammonium persulfate (APS) and ferric chloride (FeCl3) respectively. The exact mechanism of formation of these nanostructures is not very well understood, but it is argued that the lamellar supramolecular of the surfactant/oxidant is the probable precursor to these nanostructures [50]. Such nanostructures are of great interest for the fabrication of fractal like microstructures in which we have control formation on the features in the microscale as well as the ones in the nanoscale. Electrodeposition of aqueous solutions containing CTAB, CTAB and PY monomers, CTAB and Py monomers as well as APS, were identified as good candidates for electrodeposition of nanostructures on the surface of electrodes. The fractal like nature of the final device was also enhanced by using a prefabricated C-MEMS array of 100 μm high carbon posts as a substrate electrode. The combination of electrodeposition and photolithography techniques spans more than three orders of magnitude of control over the dimension of the structures being fabricated.
3.2.2. Materials and Methods All chemicals used in the experiments were purchased from sigma Aldrich. A Gamry© FAS1 potentiostat was used for all electrochemical experiments. All experiments were done using a Ag/AgCl reference electrode. A gold plated glass electrode 8 to 12 times larger than the working electrode was used as a counter electrode. A 23 mM CTAB solution was prepared by vigorously stirring CTAB crystals into deionized water for a period of 10 hours at room temperature, followed by 10 minutes of stirring at 40oC. The solution was always used within five days of preparation. Pyrrole monomer solution was used as received, stored in a freezer (≈-10oC), and stirred into the premixed solution of 70
CTAB for 1 hour at room temperature. A 15μmM Py concentration was typically used. The solution was used within two days of mixing. APS was added to the premixed CTAB and PPy and typically stirred at room temperature for 1 complete hour before being used in the experiment. The APS solution was used within three hours of preparation. Scanning electron microscopy (SEM) images were obtained using a Hitachi S-4700-2 field-emission scanning electron located at UCI’s INRF. Energy dispersive spectroscopy results (EDS) were obtained using a Carl Zeiss Ultra electron microscope located in the Carl Zeiss Center of Excellence at UCI's CALIT2. A beam voltage of 10KV was used for the EDS and most SEM images.
3.2.3. Results Electro-oxidation of CTAB A C-MEMS prepared layer of pyrolyzed SU-8 carbon was immersed in a 23 mM aqueous solution of CTAB. Electro-oxidation of CTAB created a layer of a material that have very crystalline characteristics. The layer of material has not been fully characterized but what is known is that it is quite similar to bromide crystals reported in the literature. We will refer to the process as electrochemical crystallization similar to what is seen in ref. .The results of electrochemical crystallization on carbon are shown in Figure 3-10.
71
(a)
(b)
Figure 3-10: SEM image of Electrochemical Crystallization layer from the electrooxidation of CTAB. The same magnification is used for both samples (1.5K). Cyclic Voltammetry between 0 and 2V at 20mVps for 4 cycles (13 min) is shown in (a). A potentiostatic deposition at 1.5V for 10 min. is shown in (b).
When the CV deposition method was used on an array of C-MEMS carbon microstructures fabricated in accordance with [5], the resulting structures in Figure 3-11.
72
(a)
(b)
(c)
(d))
Figure 3-11: SEM images at increasing magnification (130x, 300x, 1.5Kx, 3Kx), for (a, b, c, d) showing the results of electro crystallization from a CTAB solution using CV anodic deposition technique. The density of electrochemically deposited crystal is smaller in the CV deposited sample when compared to the potentiostatic deposition. The details of the electrochemistry will be discussed below. After electrodeposition, the sample is normally lifted from the electrodeposition solution and blow dried instantly. The crystal layer is loosely bound to the carbon surface at small connection lines. This explains why no crystals are found in between the posts as those have been removed during the blow drying of the electrodeposition solution away from the sample. The top of the previously circular 73
microposts is covered with little plates of crystal that are as thin as 500nm and tens of microns wide. These plates seem to have a larger deposition rate at the top edge of the posts where the electric field is normally the highest as seen in Fig 3-12. Similar field concentration have been modeled in section 5.2. The trace of the attachment of the plates onto the surface can be seen in the background of Fig. 3-12.
Figure 3-12: SEM image of the top of one of the microposts showing an increased rate of electrochemical deposition at the high field locations. In the background, the bottom surface between the post showing little bristles residues of where the crystals were previously attached to the surface, before the blow drying step.
74
Electrochemical and EDS Results
Figure 3-13: CV of CTAB on a carbon surface. The voltage was scanned from 0-2V vs. Ag/AgCl at a rate of 20 mV/s. The first cycle is shown in continuous (blue) and the second cycle is shown in crosses (red)
Figure 3-14: Chronoamperometric graph of the potentiostatic deposition of bromide crystal at 1.5 V vs. Ag/AgCl on a carbon surface.
75
Figure 3-15: CV of a 23 mM CTAB solution at 20 mV/s on Copper. First cycle is shown in blue (continuous). The second cycle is shown in red (dense crosses). The last in purple (sparse crosses)
Figure 3-16: CV of 23mM CTAB and 15mM Py solution without APS. The scan was 50 mV/s between 0-2 V. First cycle is shown in blue (continuous). The second cycle is shown in red (crosses).
76
Figure 3-17: CV of a 23 mM CTAB, 15mM Py, 15mM APS. The scan rate was 50 mV/s between 0-2 V. First cycle is shown in blue (continuous). The second cycle is shown in red (crosses).
Atomic %
C-K 87.896
N-K 2.815
O-K 0.000
Br-L 9.288
Figure 3-18: EDS results shown for the bromide crystal deposited using a 20 mV/s CV between 0-2V from a 23 mM CTAB onto a carbon surface.
77
Atomic %
C-K N-K O-K Cu-K Br-L 59.208 2.392 1.177 17.932 19.291
Figure 3-19: EDS results shown for the bromide crystal deposited using a 20 mV/s CV between 0-2 V from a 23 mM CTAB onto a copper surface.
Scanning Electron Microscopy
Figure 3-20: SEM image of bromide crystals deposited in the presence of APS and Py. The deposition solution was a 3 mM APS, 23 mM CTAB, 15 mM Py. Potentiostatic deposition at 1.5 V for 10 min was used.
78
Figure 3-21: Optical microscopy image of the sample in Fig 3-17, showing a window of approximately 1mm x 1.4mm. The outcome is very similar to floral drawing designs or typical palm tree drawings
Figure 3-22: SEM of bristles of electro and chemically oxidized CTAB and Py obtained in the presence of APS. 79
(a)
(b)
(a)
(b)
Figure 3-23: SEM image of bromide crystals codeposited in the presence of APS and Py. The deposition solution was a 15 mM APS, 23mM CTAB, 15mM Py. Potentiostatic deposition at 1.5V for 10 min was used.
(a)
(b)
Figure 3-24: SEM image of bromide crystals codeposited in the presence of APS and Py. The deposition solution was a 15 mM APS, 23 mM CTAB, 15mM Py. A 50mV/s CV between 0-2 V for 15 consecutive cycles. 80
Pyrolysis Results
Figure 3-25: SEM image of pyrolyzed bromide crystal deposited in the presence of both Py and CTAB (see Fig 319).
Figure 3-26: SEM image of carbon microposts covered with bromide crystal and PPy in a two step electrodeposition process. The first process is shown in Fig. 3-11. The second 81
process is a potentiostatic electrodeposition of PPy from a 0.15 M Py, 23 mM CTAB solution at 1.5 V for 20 min.
Figure 3-27: SEM images of pyrolyzed structures obtained by consecutive electrodeposition of bromide crystals and Py on C-MEMS carbon microstructures (see Fig. 3-21)
82
3.2.4. Discussion Bromide Crystal Electrodeposition
From the SEM images shown in Figures 3-11 and 3-12, it is clear to us that the deposited material resulting from the electro-oxidation of CTAB on carbon is a unidentified crystal that strongly resembles bromide crystals shown in the literature. The exact identification of the electrodeposited material is beyond the scope of this work, but will surely be investigated at a later time. Energy dispersive spectroscopy (EDS) was used to identify the composition of the electrochemically deposited crystal. CV based deposition was used to electro-oxidize CTAB onto a carbon surface. The EDS of the surface is shown in Fig. 3-18. The atomic ratio of Br to N is 3.3 which is not the same ratio as in CTAB (C19H42BrN or CCCCCCCCCCCCCCCC[N+](C)(C)C.[Br-] ). The ratio of C to N is 31.2. Given the layer was deposited on top of a carbon electrode, the ratio of carbon is not informative. Nevertheless the layer is surely not just CTAB (from the Br to N ratio). The general equation of the deposit would something close to C31HxBr3N. Further analysis is needed to identify the nature of this layer, but we will refer to it right for now as a bromide crystal, since non crystalline compounds of Br with similar ratios of C to N are typically either liquid or gaseous substances and do not exhibit similar faceted growth. The color of an approximately 30 μm layer of the bromide crystal is pale yellow. A 4 cycle CV at 20 mV/s is used for deposition and characterization. The first two cycles of the CV performed on an approximately 1 cm2 carbon surface is shown in Figure 3-13. The scan starts out at 0 V Vs Ag/AgCl with almost no current flowing through the 83
system. A quick and steady increase in anodic current starts at around 850mV and increases steadily with a sign of diffusion limitation occurring around 1.5V at a corresponding 700μA/cm2 oxidation current. In the backward sweep a cathodic peak is observed at 300 mV. The current at zero potential is a reductive current indication that there is a new species that was formed during the cycle and that is being reduced. The cathodic peak associated with the new species is seen to decrease in the second cycle along with the intensity of the anodic current when compared to the same potential during a previous scan. The bromide crystals formed through CV on the carbon surface is shown in Figure 3-10 (a). Curved and lamellar structures with a thickness smaller than 2 μm are deposited. The surface density of the structure thus obtained appears to be less than the ones
obtained
through
potentiometric
deposition
in
Figure
3-10
(b).
The
chronoamperometry of the electrodeposition at 1.5 V vs Ag/AgCl exhibits a strong current at the beginning 800μA/cm2 that decreases gradually and to a plateau at 50 μA/cm2. The potentiostatic deposition leads to a denser coverage of the bromide crystal. It is envisaged that the growth process of the bromide crystal film is a type of bottom growth model where the high currents at the beginning are associated with fast growth which forms the high aspect ratio crystals therefore limiting diffusion to the bottom of the surface. SEM observation show a decent level of charging glare which indicates the bromide crystal is not a great electronic conductor (if at all conductor). The bottom growth model is viable even if the bromide crystal was a completer insulator. The poor attachment to the surface as well as the high aspect ratio make it such that a strong water rinse step can get rid of most electrodeposited material. Nevertheless for the purposes of 84
using CTAB as a template for deposition of other material, the adhesion to the surface is good enough to allow a drying step then a reinsertion into another solution for a two step electrodeposition cycle. Electrochemical crystallization process has been previously observed and used in CuO2 oxide as described in Ref. [51]. That reference also describes how the presence of surfactants can lead to a dramatic variation in the end result of the crystal due to preferential adsorption of surfactant molecules on the different crystalline faces. A similar process can explain the really high aspect ratios of the electrochemically deposited bromide crystal shown in figure 3-11. Once nucleation on the surface of the carbon occurs, the crystal starts forming from CTAB, further growth is completely dictated by the high adsorption of CTAB, present in the solution, itself onto the crystal plate parallel to the small plate surfaces. This leads to high aspect ratio growth. Evidence of the nucleation process that is being referred to is shown in the background of Figure 312, where strong fluid flow had washed away the electrodeposited plates and the attachments to the surface (nucleation sites) were the only things that was left from the structures. Since this preferential growth is dependant on the surface concentration of the surfactant molecule [51], a potentiostatic deposition which depletes the concentration of CTAB at the bottom of high aspect ratio plates, locally inhibits the preferential allow for renucleation to occur at different places and therefore justifies the denser packing shown in 3-10(b). This same effect explains why the highest aspect ratio microplates are obtained in location where the migration effects greatly enhance the concentration of CTAB by the mere increase in the localized electric field.( see section 5.2).
85
When CTAB is electro-oxidized on copper surface a different CV shows quite a different behavior of the system. an oxidation current is noticed to rise from the beginning of the forward sweep from 0-2 V (see Fig 3-15). The slope of this current is constant all the way to 2V. The return sweep does not show the cathodic peak at 300 mV that was observed in the case of the CTAB on carbon. There is return to an almost zero current at 0 V for further cycle. The results of the CV indicate that the electro-oxidation of CTAB on copper generates products of a different nature than the ones obtained on carbon. Effectively, the EDS results show a completely different atomic % distribution (see Fig. 3-19). The Br to N ratio in this case was 8.07 compared to the 3.3 for the carbon deposited sample. This shows a big influence of the electrode surface on the formation of the bromide crystal with high aspect ratio plates. Those plates were not observed when electrodeposition was performed on copper. The Br to N ratio is very different from 1, which means that the bromide crystal is definitely not CTAB.
Electro-Oxidation of a Mixture of CTAB and Py When Py was added to the CTAB solution at a 15mM concentration, all forms of microstructure formation was repressed. The deposition is very similar to the electropolymerization of Py in a 0.1 M NaDBS aqueous solution although with potential shift of almost 300mV. The CV of the CTAB and Py solution is shown in Fig. 3-16. no cathodic element is observed in the sweep. A strong reduction of current between successive cycles is observed. The effect of hydrolysis which would normally appear at
86
almost 1V Vs Ag/AgCl is not directly distinguishable. Carbon’s wide stability window for hydrolysis is probably responsible for this effect.
Electro-Oxidation of a Mixture of CTAB and Py in the presence of APS When a solution of CTAB and Py is augmented by APS, electrodeposition start displaying intricate crystalline structures depending on the concentration of APS, and the electrochemical deposition technique used. Figure 3-20 shows the results
of
potentiostatic deposition in a Py CTAB mixture with a low concentration of APS (3mM) The carbon surface appears to be covered with submicron thick bristles, with curved arcs. And leaf like shapes. The length of the bristles can go up to 60 μm. Bristles of exactly similar shapes were reported in passive chemical oxidation of Py on a Mica surface in the presence of CTAB [52]. These structure are not conserved through a typical pyrolysis process strongly contradicts the assumption set for by the authors of [52] which believes them to be PPY. Given that these structures can be obtained by both electrodeposition and passive chemical deposition, we believe that they actually are indicative of the self assembled supramolecular configuration of CTAB/PY/APS precursor materials. Electrodeposition for longer times only results in denser surface coverage of similarly shaped microstructures, which are believe to be mostly CTAB. The trouble with the precursor structures in solution model is that one would expect to have structures deposit at multiple onto the surface of the carbon and therefore yield a thick 3D structure. If that actually happens, these structures would be very loosely bound and be washed away in the processing steps. Some of those bristles are actually seen poking out around 5 μm off 87
the surfaces in areas where they are densely deposited (see Fig. 3-21). The bristles are believed to be mostly made out of electro-oxidized CTAB for multiple reasons: -
The bristles have a week attachment to the surface and could be rinsed out with a moderate force flow which does not indicate strong covalent bonding to the surface of the carbon substrate. PPy on the hand is normally quite strongly attached to the surface and would easily resist a washing step.
-
The show sign of charging in SEM images indicating a rather low electronic conductivity atypical of PPY.
-
They deform under SEM observation which doesn’t happen for PPy.
-
The bristles’ electrodeposition CV (Fig. 3-17) is very similar to that of CTAB alone in solution ( Fig 3-13)
-
The bristles cannot survive the pyrolysis process (see Fig, 3-25), which is usually survived by PPy.
Note simply taking a sample droplet of the mixture solution and observing it on a glass slide under the microscope does not reveal similar structure’s presence in solution. Chemical means alone is thus not capable of forming these small structures with the few hours of testing (5h). We conclude that Ref. [52] improperly identified these structures as made out of PPy while there is evidence to the contrary. More assurance can be given by other more specialized material characterization techniques.
88
When the concentration of APS is increased longer range straight fibers are obtained under potentiostatic deposition (Fig. 3-23). 300 μ m submicron long straight fibers seem to deposit in as stacking like fashion. When a CV is used, the long range straight structure transformed into crooked small aggregations that seem to have in them as many bend as the number of cycle reversal (15 cycles for the samples in figure 3-24). Still the deposition seems rather 2 dimensional and the resulting material does not survive the pyrolysis process (see Fig. 3-25).
Combination of C-MEMS with bromide crystal deposition and PPY deposition The sample from figure 3-12 was taken and put in a PPy electrodeposition solution. Electrodeposition was performed for 20 min at 1.5V, long enough to get a coating layer of Py all around the C-MEMS samples covered by both the bromide crystals and PPy see Fig. 3-26. After pyrolysis in a non reactive environment the remaining structures are shown in Fig 3-27. It appears that the bromide crystal evaporated in the process as witnessed by the small cavities that are left behind. The PPy on the other hand is carbonized and the final constructions show that it is possible to obtain high surface area fractal like C-MEMS structures with a variety o features on different dimensional scales.
3.2.5. Conclusion
In this chapter, we have demonstrated the electrochemical crystallization of a bromide crystal on top C-MEMS microposts. Potentiostatic and CV based electrodeposition 89
techniques were used to deposit a bromide crystal with aspect ratio larger than 20:1. Similar structures have been believed to be PPy in Ref [52]. EDS and SEM has been used to characterize the electrodeposited bromide crystal. The combined C-MEMS and crystal structures were used as a template for further electrodeposition of PPy. The combination of photolithography and two successive electrochemical depositions was followed by a carbonization that lead to a complex glassy carbon microstructure with submicron features. The structures have a fractal like nature and the fabrication method described is a promising method for obtaining controlled multiscale carbon MEMS electrodes for applications in chemical and biological microsensors, microbatteries, and micro fuel cells where optimal control over the structure at different scales is desired.
90
Chapter 4
4. Mixed Chemical and Physical Modifications to C-MEMS 4.1.
Porous C-MEMS
In an effort to fabricate anodes that can support higher current densities, an increased surface area across which the charge transfer reaction can occur is essential. Several methods have been used in order to increase the surface area of C-MEMS Microbatteries. These methods aim towards an overall goal of creating multilevel fractal electrodes that enhance both the charge transfer through a high exchange surface area but also the current collection through a low resistance path towards metal current collectors. During the pyrolysis step, the photoresist transforms into carbon by replacing the carbon chemical bonds with nitrogen, oxygen, and hydrogen with mostly carbon carbon bonds. During the process reactive hydrocarbon gasses are formed. It is thought that if the 91
temperature ramp rate of the pyrolysis is made quick enough, the flow rate of the inert gases is insufficient to remove the reactive gas. These reactive gases (e.g. HF that are part of the decomposition products of SU-8 [7]) linger close to the microstructures and react to form other gases that leave behind micropores as seen in Fig. 4-1. Ramp rates of around 15Co/min to reach 900Co can produce satisfactory results Fig 4-1(a). This microporosity can be enhanced by using even faster rates (90C0/min) resulting in l0 to 15 micron pores shown in Fig. 4-1(b). At the slower rates some pores are observed below the surface of the microposts as seen in the purposely broken posts in Figure 4-1(c). It is not known if these pores form a communicating network.
a
b
c Figure 4-1: SEM images of Microporous C-MEMS obtained using fast pyrolysis at different ramp rates: (a) 15Co/min and b) 90Co/min from room temperature to 900Co 92
4.2.
Plasma Treatment
While fast pyrolysis can make microporous high aspect ratio C-MEMS structures, treatment with oxygen plasma can increase the surface area on the submicron scale. Surface roughening was done by subjecting the pyrolysed samples to an oxygen plasma environment. Using a moderate power of 300 Watts, under 200 mT of oxygen, for 4 minutes was enough to dramatically change the texture of the C-MEMS microstructures. Two types of indentations were observed as a result of the plasma treatment: larger indentations around 2 microns in size and smaller ridges around 200 nm in width (see Fig. 4-2).
Figure 4-2:–SEM of C-MEMS surface after plasma treatment.
93
Regular C-MEMS Plasma Treated
Figure 4-3: Nyquist plots of EIS performed on plasma treated and untreated sample. Electrochemical testing was conducted inside an argon glove box. The lithium intercalation was confirmed through multiple cycles in and out of C-MEMS anodes and results were previously reported in [53]. Electrochemical impedance spectroscopy (EIS) was used to evaluate the effective increase in surface area that can be obtained using oxygen plasma as an example. The tested samples were 2.7 microns thick and 1 cm in diameter. One was treated with oxygen plasma as previously described. The other control sample was left untreated. The C-MEMS surfaces were then placed in a teflon electrochemical cell as working electrodes. The counter and reference electrodes consisted of a lithium metal foil. The electrolyte used was a 1M lithium perchlorate in a 1:1 Ethylene Carbonate/Dimethyl Carbonate mixture. EIS was performed using a 10mV rms amplitude, from 100Khz to 1Hz. A simple Randles model was used to fit the data and the resulting parameters were extracted from Fig. 4-3.
94
Table 4-1 – Electrochemical parameters extracted from the analysis of Nyquist EIS plots.
Regular
Plasma treated
C-MEMS
C-MEMS
Rohmic (ohm)
90
102
Rct (ohm)
550
148
Io (µA/cm2)
57
216
The exchange current density, Io, was multiplied by almost 4 which can somehow reflect an effective increase in surface area by a factor of 4 (see Table 4-1). This conclusion assumes that the increase in kinetics of intercalation was solely due to increase in surface area. The chemical nature of the surface as well as the bulk of the material has been probably affected and a conclusive remark is hard to make. Still Plasma treatment allows faster intercalation of lithium into carbon anodes fabricated using the C-MEMS process 4.2.1. Conclusion
95
Fabrication of higher surface area C-MEMS structures has been demonstrated. Oxygen plasma treatment of C-MEMS has lead to a more facile lithium ion charge transfer therefore enabling higher power density microbatteries.
4.3.
C-MEMS Mixed with Nanostructures
C-MEMS can be fabricated with precursor materials already containing nanostructures such as nanoparticles, nanotubes and nanofibers. Several strategies were followed to obtain a final structure with nanofeatures. The process consisted of mixing an SU-8 solution with a nanostructure and then doing regular SU-8 photolithography followed by pyrolysis. The SU-8 photolithography steps had to be adapted to fit the needs of the mixed material which often resulted in longer exposure and bake times. The same method
C-MEMS with Carbon Nanofibers
100mg of carbon nanofibers obtained from a commercial provider (Hawaii industries, inc.) were mixed with 15.2 g of SU-8 100 and Experimental NO2 SU-8 provided by MicroChem, MA. The carbon fibers were a broad mixture of multi walled carbon nanotubes in a variety of shape shown in Figure 4-4 (straight fibers, spiral like, bent, etc.). 96
Figure 4-4: SEM image showing the morphology of CNF obtained from Hawaii industries The nanofibers were grown in a sequential oxidation reduction method using CO gas as the hydrocarbon source in the presence of iron as a catalyst. The mixing was done through manual stirring using a small glass rod. The resulting mixture is more viscous than regular SU-8 100. An exposure dose of 900 mJ was used in order to compensate for the optical absorption of the carbon nanofibers. Multiple experiments with various baking times were performed where the carbon nanofibers did not survive the pyrolysis step. The typical results looked like what can be seen in Figure 4-5
97
(a)
(b)
Figure 4-5: SEM image showing the top of an SU-8 CNF micropost before (a) and after (b) pyrolysis. The image shows the oxidation/removal of CNF from the surface of the carbon microposts. The critical step in the process was a post development bake at 95oC for half an hour. This added step made sure that the traces of developer and Isopropanol had been completely evaporated from the crosslinked SU-8 matrix before high temperatures were attained. It is speculated that the presence of remnant solvents during the pyrolysis step was the major causing of the etching/removal of the CNF from the surface of the CNF SU-8 C-MEMS. The results are show in Figure 4-6. It is observed in close up views that no growth of CNF fibers occurs but rather the same CNF are conserved. The matching was done using Photoshop® on SEM images with the same magnification.
98
Figure 4-6: SEM image showing an SU-8 CNF micropost before (a) and after (b) pyrolysis. The CNF have survived the CNF process without damage much etching. The shrinkage between the two posts is evident. Both images were taken at a 900x magnification. The substrate in the carbon sample is cracked because it was thick than 35 μm of SU-8.
It was earlier believed that CNF grow during the pyrolysis process by a local chemical vapor deposition process, but multiple testing in an effort to repeat that experiment failed. The density of CNF at the surface of microposts depends on the initial mixture of CNF SU-8. The shrinkage in SU-8 makes it look as the surface density of the CNF increases but this was in fact optical illusion because the whole micropost structure shrank by a considerable amount while the CNF was not affected by the shrinkage. When higher concentration of CNF (200mg/ 15 mg of NO2 SU-8 100) was used, the results looked like in Figure 4-7.
99
Figure 4-7: SEM image of CNF covered C-MEMS microposts using twice the density of the CNF in NO2 SU-8 100 as those shown in Figure 4-5.
C-MEMS with Catalyst Nanoparticles
The most common model followed for the chemical vapor deposition growth of carbon nanofibers is the one described in depth in [54], (see Fig 4-8). The basic principle is the decomposition of hydrocarbon gas at the surface of the metal nanoparticle, followed by a formation of a metal carbide phase. When the metal carbide phase saturates with carbon, precipitation in the form of graphite is observed on the substrate side of the metal nanoparticle.
100
Figure 4-8: Schematic representation of carbon nanofiber catalytic growth mechanism from a hydrocarbon gas as described in [54].
It appears that such a process is quite susceptible to contamination from external impurities and more specifically all the components present in SU-8 photoresist. The common catalyst materials used in CVD CNF growth are Fe, Ni, Co. Nanoparticles of these catalyst materials were mixed with SU-8 NO2. The samples were later pyrolyzed in both a forming gas environment and a methane atmosphere at 900oC. None of the experiment lead to any substantial production of CNF (see Figure 4-9)
101
(a)
(b)
Figure 4-9: SEM images of an SU-8 Cobalt microposts before (a) and after (b) pyrolysis. The images were taken at 800x and 1500x for (a) and (b) respectively. The . Instead various low density unidentified fibers of the different metal/metal oxide were observed See Fig 4-10. Such experiments need to be done with a more controlled polymer and proof of concept on that level needs to be done without the complexity of a photoresist mixture where multiple ingredients can “poison” the catalyst particles and inhibit the formation of CNF.
102
Figure 4-10: SEM image of nanowires grown unpredictably on the edge of a Cobalt nanoparticle agglomeration.
4.4.
C-MEMS with in situ Grown Nanofibers
In collaboration with Michigan Tech University MTU (professor Yoke Khin Yap), CMEMS structures were fabricated at UCI and sent to MTU for further coating with a layer of carbon nanotubes. MTU used a pulsed laser deposition technique to first deposit a thin layer of Alumina (acting as a diffusion barrier and a substrate), then a layer of Ni catalyst nanoparticles was deposited. This served as the basis for the catalytic deposition of CNT on top of the posts. The density of deposition was not appreciable and no further results came out of the collaboration (see Fig. 4-11). The presence of the Alumina and Nickel catalyst made it hard to decouple the effects of these two layers on observed electrochemical behavior of CNT C-MEMS. 103
(a) (b) Figure 4-11: SEM images of CNT deposited on top of C-MEMS micropost (a). Low density coverage is observed in the close up 13Kx in (b).
104
Chapter 5
5. C-MEMS Fabricated Lithium Ion Anodes: Testing and Modeling
5.1.
Three
Dimensional
C-MEMS
Microbattery
Anodes from Pyrolyzed Freestanding SU-8 Polymer Films: Fabrication Process and Testing of Lithium Ion Intercalation Properties
105
5.1.1.
Introduction
5.1.2.
Materials and Methods
All photolithography, pyrolysis and SEM procedures were conducted at the Integrated NanoResearch Facility (INRF) located on UC Irvine campus. This facility is class 10000 clean room facility with dedicated photolithography areas that operate at class 1000. The SEMs were performed using Hitachi S-4700-2 field-emission scanning electron microscope (FESEM). Nano SU-8 100 (and other viscosities) was purchased from MicroChem ©,MA. PGMEA (>99.5%) was purchased from Sigma Aldrich and used as a developer for SU-8. Polystyrene Petri dishes were purchase from Sigma Aldrich and used as substrates for SU-8 spin coating. Cold rolled (CR) steel 2 in x 2in square plates: 4 gauge (≈1/4 in.)10 gauge (≈1/8 in.), and 16 gauge (≈1/16 in.), density of 8.03g/cc and were purchased from Industrial Metal Supply, Irvine, CA. The paraffinic oil Fisherbrand 19 vacuum oil, was purchased from Fischer Scientific, PA and was used as lubricant between SU-8 and cover plates. All electrochemical tests were performed using a VMP3 potentiostat purchased from AMETEK Princeton Applied Research, TN. LiClO4 and Ethylene Carbonate (EC), DiMethyl Carbonate, and lithium film were all purchase from Sigma Aldrich for use in electrochemical testing. Military grade Viton o-rings were purchased from McMaster Carr, IL. A 3.5 in. quartz tube flow through furnace was used for pyrolysis and hard baking of the SU-8 free standing films. Gas flow rates in the tube were fixed at 2000 standard cubic centimeters per minute (SCCM). The same flow rate was used for both Nitrogen and 106
forming gas (95%N2, 5% H2), which results in an average flow velocity of approximately 0.5cm/s. all mass measurements were done using a Toledo AB54-5 METER scale with 0.1 mg accuracy. Layer thicknesses were measured using SEM. Layer dimensions of layers were measured using a standard ruler with 1 mm accuracy.
Fabrication of Freestanding SU-8 Films Fabrication of SU-8 photoresist films of various thicknesses was performed using the Rapid Freestanding SU-8 techniques explained in detail in section 2.1. In brief, SU-8 is first spin coated on a polystyrene (PS) substrate. The sample is then soft baked at 95oC, followed by a UV light flood exposure (a mask is used for the grid samples). The samples are then post exposure baked at 95oC. The SU-8 film is then peeled from the flexible PS substrate using regular wafer tweezers. The films detach easily from the substrate because of the low adhesion between PS and SU-8. Samples are then cut into 3cm x 3cm square pieces. Some samples were weighed in order to establish the yield from the general pyrolysis process. SU-8 film samples were fabricated with thicknesses varying between 25 and 150 μm.
107
Table 5-1: Process parameters shown for the different thickness of SU-8 films. Pre-spin and spin times were 10 sec. and 30 sec. respectively.
Nominal
Type
of Pre-spin
Thickness SU-8
Softbake
RPM/
Time
Spin RPM
95oC
Exposure Time Post Exposure at Using 6mw/cm2 Time at 95oC UV Source
25 μm
SU-8 25
500/2000
20 min.
25 sec.
5 min.
50 μm
SU-8 50
500/2000
25 min.
50 sec.
6 min.
75 μm
SU-8 50
500/1600
35 min.
60 sec.
10 min.
100 μm
SU-8 100
500/3000
45 min.
1min. 10 sec.
13 min.
125 μm
SU-8 100
500/2500
60 min.
1 min. 20sec
15 min.
150 μm
SU-8 100
500/1750
70 min.
1 min. 25 sec
16 min.
C-MEMS Pyrolysis Process The layers of SU-8 obtained using FSR are then placed on a quartz flat boat surface and covered with a plate made of quartz, thermally grown SiO2, or cold rolled steel. The quartz boat is the placed inside a 3.5 in. quartz tube flow through furnace operating under nitrogen gas atmosphere. The process normally starts with a 30 minute hard bake at 200oC. For the fast ramp sample, the hard bake is bypassed. This is normally followed by 108
heating until 900oC at an average rate of 10oC/min. The samples are kept at 900oC for one hour and then left to cool to room temperature. In the case where forming gas is used, the same steps are performed except that nitrogen is switched back at the end of the 1 hour at 900oC and is used to bring back the sample to room temperature (in approximately 8 hours).
During the pyrolysis process, the SU-8 films required a cover plate to ensure that the samples remain flat. When no cover plates were used, the samples warped into various shapes as seen in Fig. 5-1. Internal stresses can build up from different pyrolysis rate of different parts of the film (e.g. the conditions on the top surface exposed to the flowing nitrogen were different from he conditions at the bottom surface in contact with the quartz boat plate). When a cover plate was used, the sample remained relatively flat but sometimes broke into multiple small pieces.
109
Figure 5-1: Photograph of pyrolyzed SU-8 films. The flat sample to the left was obtained through process F in table 5-2. The warped sample to the right was pyrolyzed without a cover plate. Notice the shiny carbon reflective surface indicating a very smooth surface. µ Two processes would lead to a breakage of the samples. A strong adhesion of SU-8 to the cover surface, and the weight of the cover surface. Both would lead to a strong friction force between the SU-8 film and the plate. The glass transition temperature of SU-8 is approximately 64oC[24]. The heated SU-8 samples would tend to go through a glass transition phase as they are heated up and can easily tear resulting in a broken carbon film (usually two or more pieces). Various cover plate materials were used: Si/Sio2 wafers with ¼ in. steel weights, ¼ in. quartz plates, 2in. x 2in. cold rolled steel plates with three 110
different thicknesses (1/4, 1/8, 1/16 in.). The different plates were more or less successful in producing an intact flat carbon film. The results are summarized in table 5-2 Table 5-2: Summary of outcome of pyrolysis process for different cover plates and under different conditions. All initial SU-8 films were cut into 3 x 3 cm square shapes and placed on a quartz plate.
Type of Cover Plate and conditions A B C D E F G
Si/SiO2 quarter wafer weighted with 1/4 x 2 x 2 in. CR steel hard bake performed 1/4 x 2 x 2 in. quartz plate hard bake performed 1/4 x 2 x 2 in. CR steel plate hard bake performed 1/8 x 2 x 2 in. CR steel plate hard bake performed 1/16 x 2 x 2 in. CR steel plate hard bake performed 1/16 x 2 x 2 in. CR steel plate hard bake performed lubricating oil used 1/16 x 2 x 2 in. CR steel plate lubricating oil used hard bake skipped/ fast ramp used
Outcome of Pyrolysis All films shattered into very small pieces (2mm size on average) All films broke into four or more pieces All films broke in two or more pieces 20% of films came out intact 40 % of films came out intact 70% of films came out intact 90% of films came out intact
Given that the fabrication process is lengthy and tedious, it was important to identify the best process conditions for a good carbon film fabrication yield. When heated, SU-8 films adhered strongly to a to thermally grown SiO2 surface. As the samples are being heated, SU-8 first starts flowing and deforms easily under mechanical stress. After reaching 300oC [27], the SU-8 starts carbonizing and losing mass. The process continues at higher temperature with the associated shrinkage. If the adhesion to the cover plate is too strong, the SU-8 layer is not allowed to shrink in one single piece, and the sheet stress 111
becomes overwhelming and leads to breaking the layer into multiple pieces. This occurs strongly for SiO2 (entry A in Table 5-2) and to a lesser extent for a quartz plate (entry B in table 5-2). Better results are obtained using cold rolled steel plates. The lighter thinner steel plates (1/16 in.) have significantly better yield than the thicker plates. Better results are obtained when one drop of vacuum oil is used as a lubricant between the steel and the SU-8 films, and also between SU-8 film and the bottom quartz plate. The best process yield (90%) is obtained when, in conjunction with using the paraffinic oil lubricant, the hard bake step is skipped (in a fast ramp pyrolysis). It is possible, during that step, that the SU-8 adheres strongly to the cover plates. When fast ramp is followed without the use of oil, no improvement in process yield is observed. Processes F and G from table 5-2 were used for samples that were electrochemically tested. Processes B, D, and E were used for measuring the yield of the pyrolysis with respect to mass (ratio of the mass of carbon film and the SU-8 film).
Electrochemical Testing All lithium intercalation electrochemical testing was performed in a glove box under argon atmosphere. A two electrode system using Li metal film as both reference and Counter electrode was used. A poly tetra fluoroethylene (PTFE) cell was specially machined to accommodate the testing of pyrolyzed SU-8 films. A diagram of the initial design can be found in the Appendix. The later designs were quite simplified. The sample under test was separated from the lithium film using a 3mm thick Viton® o-ring that exposed a disk of 0.9 cm diameter (area of 0.636 cm2). The 3mm separation was filled 112
with liquid electrolyte through a whole in the lithium metal. 1M LiClO4 in a mixture of EC:DMC in (1:1 volume ratio of Ethylene Carbonate and DiMethyl Carbonate) was used as the electrolyte. A metal copper foil attached to a glass plate was used as a current collector from the anode side (see Figure 5-2). The metal foil was contacted to the carbon plate by applying pressure between concentric o-rings (using a thumb screw and wing nut system). Lithium foil was cut and the surface exposed to the solution was polished with a coarse sand paper so as to expose a fresh Li metal surface. Li was then perforated in the middle with a 2mm diameter drill bit. The assembly was secured and then electrolyte was introduced using a glass pipette. Electric tape was then used to seal off the top of the electrolyte cavity in order to prevent electrolyte evaporation during the length of the test. The copper and Li films were contact using stainless steel alligator clips. The full assembly is shown in Fig. 5-3.
Figure 5-2: The components of the cell assembly: perforated lithium film on inverted PTFE cell (left). Wing nut used to put pressure on the assembly (left). Copper on glass used for current collection (right). Insulating plastic sheet used as an electric short guard. Viton O-ring used as spacer between the Li foil and the carbon films.
113
Figure 5-3: Final assembly of the PTFE testing cell used in the electrochemical testing. The EC/DMC 1M LiClO4 electrolyte was introduce using a pipette from the top whole that is later covered with electric tape to avoid evaporation effect during long term testing. The main testing routine that was performed consisted of first an initial lithium insertion and SEI formation at a rate of 100 μA (i.e. 160 μA/cm2) down to 50 mV vs Li/Li+. Lithium deintercalation is then done at the same rate of 100 μA until a potential of 2V vs Li/Li+ is reached. Subsequently, all intercalations are done at the same base rate of 100 μA, and deintercalations are done at the rates of (5x, 10x, 30x) of that base rate corresponding to (0.5 mA, 1 mA, 3 mA). Each time 50mV is reached, an Electrochemical 114
Impedance Spectroscopy (EIS) is performed. The EIS is done at a forced 50mV Vs Li/Li+. Potentiostatically. A 5 minute stabilization period precedes the acquisition. The EIS scans from 200 KHz to 0.1 Hz using a 10mV peak to peak.
C-MEMS Film Metering Each sample intended for electrochemical tested is weighed after pyrolysis to the nearest 0.1mg. The dimensions of the film are recorded using a regular ruler with a 1mm accuracy. After electrochemical testing, the thickness of each sample is determined by mounting a portion of the sheet in a vertical fashion on carbon tape and looking at the cross section. The SEM stage is rotated in such a way as a view parallel to the film is obtained. The thickness is measured at that point. The stage orientation needs to be adjusted for each sample measured because of the variation between mounting positions. Although all SU-8 Films are initially cut into 3 x 3 cm squares, the carbon samples were not always square. Shrinkage of the SU-8 as it is pyrolyzing is hard to control. Ideally, a frictionless surface, non reactive, and can withstand high temperatures.
5.1.3. Results and Discussion
C-MEMS Pyrolysis Gravimetric Yield The carbon films obtained from the C-MEMS pyrolysis of SU-8 have different morphologies. The films were strong enough to be handled easily with tweezers but displayed the typical brittleness of a carbon material if excessive force was applied. The 115
gravimetric yield of the pyrolysis process was evaluated for films thicknesses ranging from 50 to 150 μm.
Gravimetric Yield of Pyrolysis for Different Thicknesses of SU-8 Films and Different Cover Plates 28
Percent Yield (%).
26
uncovered
24
covered 1/4 in. quartz covered 1/16 in. steel covered 1/8 in. steel
22 20 18 16 14 12 10 40
60
80
100
120
140
160
Thickness of Original SU-8 Film (microns)
Figure 5-4: Gravimetric yield of pyrolysis of SU-8 film for films ranging from 50 to 150 μm. Different cover plates were tested. Pyrolysis conditions were kept constant for all samples (hard bake at 200oC, 1hour at 900oC under N2). Each series with different thickness was processed in the same furnace run. Each point on the graph represents a single measurement. The gravimetric yield of the pyrolysis process is shown in figure 5-4. While all covered surfaces seem to be lumped in the same range (24%-27%) for all thicknesses tested, the uncovered samples had a substantially smaller yield closer to 21.5%. Although the differences are small, it is indicated that there is a positive relation between the weight of the cover plate and the yield. An possible explanation could be that, the heavier plate exerts a stronger force onto the surface of the film and therefore presents a higher 116
resistance to the outgasing of pyrolysis products. The higher resistance to the outgasing means that the higher vapor pressures are required for the gases to escape the film. This exerts a thermodynamic pressure on the carbonization reaction towards forming more carbon, given that the pressure of the products in the carbonization is higher. The effect of weight on pyrolysis is more pronounced for the thinner layers where the ratio of the surface to the total volume of the layer is quite larger. The heavier 1/8 in. of steel cover resulted in the highest yield for all thicknesses, followed by the 1/6 in. steel then the quartz plate. For a completely uncovered pyrolysis of SU-8 films, the resistance to outgasing is null (from the top surface at least), this results in the smaller yield of 21.5% for a 50 μm layer compared to the 26.28% for the same thickness layer covered by the 1/8 in. steel. It is important to note that all samples that were covered showed a glossy mirror like finish, while the uncovered samples had a matte less reflective finish. If it is assumed that the layer of SU-8 is thin enough that the temperature is the same all throughout the layer and that the preferred outgassing direction is through the thickness of the layer and not through its edges (for obvious path of least resistance reasons), it follows that the gas flow rate across the surface of the film during pyrolysis is proportional to the total volume of the film. Since the area of the film is the same for all samples, it follows that larger flow outgassing flow rates will occur for the thicker layers. Such a flow rate can easily induce microporosity at the surface and allow for more efficient outgasing to occur. The microporosity thus formed would lead to a smaller resistance to the outgasing and therefore to a smaller yield, which explains why thicker layers (with smaller surface to volume ratio) have a much smaller yield in the case of the 117
uncovered samples. Although direct SEM observation of the thicker films was not done (since the uncovered films were warped and of not usable for testing for electrochemical testing for this work), examples of the manifestation of this hypothesis were obtained in in a fast pyrolysis experiment investigating porosity.
118
(a)
(b) Figure 5-5: SEM images of 85 μm high carbon posts pyrolyzed rapidly and showing evidence of outgasing. The post shown are 435 μm high in (a), and 35 μm high in (b). It was found that larger posts with smaller surface to volume ratio seem more prone to develop microporosity as seen in Figure 5-5a where a 435 μm post is shows 20 μm pores after being pyrolyzed using a very fast ramp rate (60oC/min.). Figure 5-5b shows posts on 119
the exact same sample with much a 12.5 times smaller diameter of 35 μm which seem to have kept a surface devoid of large pores therefore presenting a larger resistance to outgasing. The outgasing during pyrolysis is expected. Fig.5-5 shows direct evidence of the process occurring while the SU-8 is still flexible. Some of the posts are seen to bloat up like a balloon, while others seem to have been able to release the pressure buildup through the creation of pores. It can be visually seen that the yield of a structure with a surface to volume ratio 12.5 times smaller is lower than the same structure having a larger surface to volume ratio. This supports the observed smaller yield for thicker structures (>100 μm ) during uncovered pyrolysis. The same difference is not visibly seen in the case of pyrolysis using cover plates because the outgasing is strongly inhibited by the pressure buildup from the top and thus the majority of outgasing occurs from the edges at rates that don’t allow for large pore formation. The presence of large pore networks can potentially inhibited by the pure mechanical pressure on the film. The formation of pores renders the structure weaker and the mechanical pressure from the top can easily collapse any pores. If the collapsed pores don’t connect to other pores, the outgasing process will be quite inefficient and the yield would therefore remain the same. The different optical properties between the covered and uncovered films is also an indicator of the level of porosity difference between the different structures
120
C-MEMS Pyrolysis Volumetric Yield
The volumetric yield of the pyrolysis process under different conditions is shown in Fig.5-6. The thicknesses of 25 µm and 75 µm are the nominal thicknesses as normally obtained using standard SU-8 recipes.
Volumetric Yiedl (%).
Volumetric Yield of C-MEMS Pyrolysis 0.3 0.25 0.2 0.15 0.1 0.05 0 25 µm under Nitrogen
25 µm under Forming Gas
75 µm under Nitrogen
75 µm fast ramp under Nitrogen
Figure 5-6: Volumetric yield of C-MEMS pyrolysis. All processes have used a paraffinic oil as a sheet lubricant (entry F in Table 5-2) to avoid cracking. The volumetric yield is hard to analyze because not only does it involves the physical effects of pressure on SU-8, it also requires a full knowledge of the mechanics of the pyrolysis process. For the same thickness, it seems that using Forming gas helps increase 121
the volumetric yield by 26%, while using a fast ramp process decreases it dramatically (by 37%). Forming gas provides a reducing environment in which can react very quickly with strong oxidants (example HF gas [7] from the decomposition of the photocatalyst generator SbF6) emanating from the pyrolysis and prevent them from etching away chemically at the rest of the surface. Fast pyrolysis on the other hand has a pronounced porosity increase effect seen in an amplified manner in Fig. 5-5 where the ramp rate was 6 times faster that the one used for our sheet samples. The volumetric yield seems to be directly reflected by the different thickness (see Figure 5-7). In fact, the change in surface of the layers seems to be almost uniform for all the samples. The samples shrank from 9 cm2 to an average of 3.2 cm2 (see Fig. 5-8) representing a shrinkage of approximately 65%. Since most surface areas shrank in a similar manner, the thickness difference directly reflected the volumetric yield. In the case of the 25 µm sheets, the thickness shrinkage was around 50% regardless whether forming gas was used or not. The 75 µm thickness under nitrogen experience only 32% shrinkage, while the fast ramp 75 µm was reduced by 60%. Volumetric yield as a value is not a very useful indicator in general except possibly when considering stress related effects. It appears in this case that the higher volumetric yield was associated with the samples that cracked the most during the pyrolysis process. This is counter intuitive, since the opposite is expected to occur. But if looking at it from a different perspective, the samples with the smaller volumetric yield being possibly less affected by adhesion to the cover plate and therefore closer to a non covered plate with from figure 5-4 show a smaller yield (by mass but there could be a
122
correlation there). Still it is not well understood why a thicker layer would have a higher
Thickness after Pyrolysis (microns)
volumetric yield.
Thickness of C-MEMS Carbon Sheets 60 50 40 30 20 10 0 25 µm under Nitrogen
25 µm under Forming Gas
75 µm under Nitrogen
75 µm fast ramp under Nitrogen
Figure 5-7: Thicknesses of C-MEMS carbon sheet.
Sheet Area (square cm).
Area's of C-MEMS Carbon Sheets 4 3.5 3 2.5 2 1.5 1 0.5 0 25 µm under Nitrogen
25 µm under Forming Gas
75 µm under Nitrogen
75 µm fast ramp under Nitrogen
Figure 5-8: The shrinkage of the surfaces of SU-8 films was almost constant at 65% of the initial 9 cm2
123
The density of the different carbon is an important parameter that affects the lithium ion electrochemical properties of the carbon. Higher density carbon can indicate higher Lithium intercalation volumetric capacity. The density of the end carbon did not seem to be affected by the atmosphere. Both the Nitrogen and Forming gas processing yielded a circa 1.8 g/cc carbon density from 25 µm films (see Fig. 5-9). The density of the thicker 75 µm was expectedly a little smaller (1.55 g/cc) in conjunction with the higher volumetric yield of that batch of samples. The highest density on of 2.3 g/cc the other hand was obtained by using the fast ramp under N2. While this density is quite high for what is normally expected of a glassy carbon (1.55-2.10[55]), and it is closer to the density of pure graphite but the standard deviations are quite large for this batch of samples. The trend is clear though, that faster pyrolysis under a steel cover yields a
Pyrolyzed Carbon Density (g/cc).
higher density carbon.
Density of C-MEMS Carbon Sheets 3 2.5 2 1.5 1 0.5 0 25 µm under Nitrogen
25 µm under Forming Gas
75 µm under Nitrogen
75 µm fast ramp under Nitrogen
Figure 5-9: Density of C-MEMS carbon under different pyrolysis conditions 124
Density of SU-8 Layers of different thicknesses 2.5
Density (g/cc)
2 1.5 1 0.5 0 0
20
40
60
80
100
120
140
Thickness of SU-8 layer (microns)
Figure 5-10: Density variation between different SU-8 film thicknesses. The trend shows a 25 µm layer can be 60% denser than a 150 µm (2.15 vs.1.36 g/cc) The density variation between different thicknesses of carbon can also be explained by the fact that different thicknesses of spin coated SU-8 have different densities as well (see Fig.5-10), when measured after the final processing steps. It was originally assumed that after soft baking, the SU-8 layers having gotten rid of their solvent content of GBL should have the same density. This difference introduces a lot of uncertainty on the discussions related to comparing results for different SU-8 thicknesses, because it appears that the original starting material is not the same.
125
Lithium Ion Intercalation of SU-8 C-MEMS Sheets
Charge/Discharge Curves: For the purposes of illustration, an example of charge discharge curve is shown in Figure 5-11.
1
0.8
Ewe/V
0.6
0.4
0.2
40,000
50,000 time/s
60,000
Figure 5-11: A typical cycling profile of a C-MEMS sample seen here for a 12.5 µm layer processed under N2. The shown profile starts right after the first intercalation process forming the SEI Layer at 100 µA. The discharge is done successively at 100 µA, 0.5 mA, 1mA, and 3 mA. Charging is always performed at 100 µA (160 µA/cm2).
Electrochemical Testing Results The samples tested for lithium ion intercalation were the ones tested under different atmospheres described in figure 5-6. Cycling was performed between the limits of 50 mV 126
and 1V vs. Li/Li+ electrode. The reason for testing wihtin this range will be explained later (See Section: Capacity of C-MEMS). When same thickness SU-8 layers were pyrolyzed under different processing conditions, the resulting carbon sheets carbonized into different thicknesses as seen in Fig. 5-7. The various electrochemical parameters that follow will be expressed as a function of the different types of samples used in the testing. The various graphs that follow summarize the different results.
In te rc a la tio n C a p a c ity (m A h ).
Capacity of C-MEMS Sheets at 0.1 mA Discharge Rate 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 12.5 µm - Nitrogen 13.5 µm - Forming 51 µm - Nitrogen 28 µm - fast ramp Gas Nitrogen
Figure 5-12: The capacity at 100 µA (or 160 µA /cm2) of the different C-MEMS carbon layers is shown for the cycle that follows the first intercalation. .
127
S p ecific C ap acity (m A h /g ).
Specific Capacity of C-MEMS Sheets 200 150 100 50 0 12.5 µm Nitrogen
13.5 µm Forming Gas
51 µm - Nitrogen 28 µm - fast ramp - Nitrogen
Figure 5-13: Specific capacity of C-MEMS Sheets for different conditions.
Volumetric Capacity (Ah/l).
Volumetric Capacity of C-MEMS Sheets 450 400 350 300 250 200 150 100 50 0 12.5 µm Nitrogen
13.5 µm Forming Gas
51 µm Nitrogen
28 µm - fast ramp - Nitrogen
Figure 5-14: The volumetric capacity of C-MEMS sheets for different conditions 128
Irreversible Capacity (%).
Irreversible Capacity of C-MEMS Sheets 100 90 80 70 60 50 40 30 20 10 0 12.5 µm Nitrogen
13.5 µm Forming Gas
51 µm Nitrogen
28 µm - fast ramp - Nitrogen
Figure 5-15: The irreversible capacity of C-MEMS sheets for different conditions
Capacity as a Function of Discharge Rate 0.8
Capacity (mAh).
0.7 0.6
12.5 µm - Nitrogen
0.5
13.5 µm - Forming Gas
0.4
51 µm - Nitrogen
0.3
28 µm - fast ramp Nitrogen
0.2 0.1 0 0
1
2
3
Discharge Rate (mA)
4
Figure 5-16: Capacity as a function of discharge rate for different C-MEMS sheets with an area of 0.636 cm2. 129
Volumetric capacity at 3mA (Ah/l).
Volumteric Capacity of C-MEMS Sheets at 3 mA Discharge Rate 200 180 160 140 120 100 80 60 40 20 0 12.5 µm Nitrogen
13.5 µm Forming Gas
51 µm Nitrogen
28 µm - fast ramp - Nitrogen
Figure 5-17: The volumetric capacity of C-MEMS sheets at a 3 mA discharge current equivalent to a current density of 4.7mA/cm2
130
90
% capacity loss between 0.1mA discharge and 3 mA discharge
80 70
Percent Loss
60 50 40 30 20 10 0 12.5 µm Nitrogen
13.5 µm Forming Gas
51 µm Nitrogen
28 µm - fast ramp - Nitrogen
Figure 5-18: Percent capacity loss between discharges at 0.1 mA and 3mA for 0.636 cm2 C-MEMS sheets. The percent loss is calculated using 100x(
[email protected] Cap@3mA)/Cap@3mA.
131
Figure 5-19: Sample output from an EIS test. Zsimpwin program was used for analysis. The frequency is shown on each point. 132
- Z '', ohm
-4
-2
0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
32
34
36
38
40
50
135k
91.3k
55
60
200k 200k 200k
61.7k
41.7k
28.1k
28.1k
19k
65
19k
12.8k
70
12.8k
8.69k
8.69k
75
5.87k
80
5.87k
3.96k
85
3.96k
Z ', ohm
90
2.68k
2.68k
95
100
1.81k
1.81k
X-25-1-a-first.txt Model : Unknown Wgt : Modulus
1.22k
1.22k
105
826
115
35.9
172 116
255
376
116 53
376 255
557
110
557
826
125
481m 710m
120
3.42
1.05
326m 481m
148m
130
100m
(a)
(b)
Figure 5-20: Chosen constant phase element model for curve fitting of EIS data (a). The model is a modified Randles cell (b) with a constant phase element [Q] to substitute for the Cdl and a Warburg and Rct in series impedance to substitute for Rs.
EIS Nyquist Plots 25 20
-Im(Z)/Ohm
15 10 5 0 -5 60
80
100
120
Re(Z)/Ohm
12.5 µm - Nitrogen 13.5 µm - Forming Gas 51 µm - Nitrogen 28 µm - fast ramp - Nitrogen Figure 5-21: EIS Nyquist plots at 50mV vs. Li/Li+ for different C-MEMS sheets. 133
Table 5-3: Zsimpwin extracted parameters from the EIS after the first cycle. R ohmic Y0 n Rct (ohms x sq. cm) (Sx s^n)/sq.cm dimensionless (ohm x sq. cm) 33.95763 1.30377E-05 0.798175 35.4729 2.493505978 3.84552E-06 0.027675787 8.783480834
W (Sxs^-0.5)/ sq.cm 0.130766509 0.018023081
Cdl adjusted F/sq. cm 1.851E-06 5.467E-07
12.5 µm
- Nitrogen
Av. SD
13.5 µm
- Forming Gas
Av. SD
37.28232 0.863462233
1.25267E-05 3.08637E-06
0.8037 0.026445794
36.86256 11.3329418
0.099819182 0.012263188
2.00702E-06 1.05071E-06
51 µm
- Nitrogen
Av. SD
31.88586 1.686449673
3.25079E-05 2.0157E-05
0.7152 0.084994235
26.44488 3.993512826
0.142177673 0.025960603
1.692E-06 3.390E-07
28 µm
- fast ramp - Nitrogen
Av. SD
35.01816 4.83177534
1.5153E-05 7.24479E-06
0.795466667 0.017438559
33.61048 6.845382173
0.111006289 0.029498113
2.1335E-06 1.0011E-06
Charge Transfer Resistance of C-MEMS Sheets
Rct (ohm x sq. cm).
60 50 40 30 20 10 0 12.5 µm Nitrogen
13.5 µm Forming Gas
51 µm Nitrogen
28 µm - fast ramp - Nitrogen
Figure 5-22: Charge transfer resistance for different C-MEMS sheets at 50mV Vs Li/Li+
134
Current Density (A/sq.cm).
Exchange Current Density of Li intercalation in C-MEMS Layers 3.000E-03 2.500E-03 2.000E-03 1.500E-03 1.000E-03 5.000E-04 0.000E+00 12.5 µm Nitrogen
13.5 µm Forming Gas
51 µm Nitrogen
28 µm - fast ramp Nitrogen
Figure 5-23: Exchange current density for the different C-MEMS sheets calculated from RCT at 50 mV vs Li/Li+
percent change (%)
Change in Cdl Between EIS post-1st Cycle and post-3rd Cycle 90 80 70 60 50 40 30 20 10 0 12.5 µm Nitrogen
13.5 µm Forming Gas
51 µm Nitrogen
28 µm - fast ramp - Nitrogen
Figure 5-24: Change in the double layer capacitance with cycling
135
Diffusion Coefficient (sq. cm/s)
Li Diffusion Coefficient in C-MEMS Layers 2.010E-11 1.510E-11 1.010E-11 5.100E-12 1.000E-13 12.5 µm Nitrogen
13.5 µm Forming Gas
51 µm Nitrogen
28 µm - fast ramp Nitrogen
Figure 5-25: Diffusion coefficient of lithium in C-MEMS sheets EIS Nyquist Plots
25
-Im(Z)/Ohm
20 15 10 5 0 -5 -10
60
80
100
120
Re(Z)/Ohm
12.5 µm - Nitrogen Li vs. Li at OCV Figure 5-26: EIS Nyquist plots of Li film and a 12.5 μm carbon Sheet
136
Discussion of Results Capacity of C-MEMS Sheets
For the purposes of electrochemical testing, 4 samples were used for each point representing a different type of C-MEMS carbon sheet. In a few cases, the sample had to be dropped because during the cell setup process, the films were shattered by an excessive clamping force between the copper sheet and the o-ring. The cycling was done between 50 mV and 1V vs. Li/Li+. The lower bound was chosen at 50mV because earlier tests to 0V showed readiness towards lithium plating at small currents. The upper bound was chosen at 1V because at higher potential for thicker films, sharp fluctuations in the charge discharge curve occurred, indicating quick mechanical changes in the structure, which we believe are due to mechanical failure due to the strongly constrained edges of the 0.9cm diameter area of sample film under test. A contact promotion layer can be used between the copper and the carbon to mitigate the effect of any structural change of the carbon on the charge/discharge curves. For our purposes, we preferred to limit the cycling to 1V where the structural change of the carbon is below failure mode. The difference between the capacity for 1V and that for 2V was measured to be less than 10%. The volumetric capacity and specific capacity thus obtained are conservative because the charge times were approximately 2 h for the thin sheets (≈13 µm) and 6h for the 51 µm and 28 µm sheets. Typical capacities for batteries are evaluated at a 24h discharge rate. This applies to porous electrodes made out of packed carbon particles and where the average current density per surface area is hard to determine. The fact that we are dealing with flat carbon sheets [56] allowed us to fix the current density at the surface 137
of the electrodes and follow an analysis that is more similar to typical electrochemical testing done using surface area metrics instead of C rates used in batteries (though the conversion is quite easy between the two). The effect of the side diffusion of lithium into the area of the carbon sheet that is not directly exposed to the solution is assumed to be negligible at the time scales of our experiment and the usual diffusion rates of Li. Figure 5-12 shows the total capacities of the different C-MEMS films. The effect of using forming gas on the capacity is seen to be negligible. Using the fast ramp pyrolysis under N2 seems to dramatically enhance the capacity of the carbon. In fact, when looking at the specific capacity plots (Fig. 5-13), the fast ramp processed sample has the highest value of 175 mAh/g. This result is quite smaller than the theoretical capacity of graphite 372 mAh/g [57] and of reported pyrolyzed epoxy novolac resins 590mAh/g [58].. The specific capacity of the 12.5 µm and the 51 µm sheets processed in the standard manner (process F in Table 5-2), was identical at approximately 127 mAh/g. This result was rather surprising because the thicker layer was originally expected to have a smaller capacity. When looking at conventional anode fabrication, an optimal size for graphite particles used is approximately 10 µm particles [59]. It was expected that thicker sheets would have a substantially lower capacity density. This indicates that lithium can easily diffuse through a thickness of 51 µm in C-MEMS made anode when the intercalation rate is smaller than 160 µA/cm2. This result has important implications on the design parameters that can be used in 3D microbatteries. It indicates that for low power applications 50 µm sheets is an acceptable thickness. Forming gas appears to have a negative effect (≈20% decrease) on the capacity density of C-MEMS sheets. This in 138
contradiction with some previously published literature that show an increase of nearly two times of the reversible capacity for a change of H/C ratio from 0.1 to 0.2 [58]. It is assumed that the use of forming gas in the presence of Hydrogen would shift the H/C ratio to higher values and that a corresponding increase in capacity would be observed. Although that increase is usually coupled with a pronounced increase in the hysteresis of the cyclic voltammetry (charge/discharge curves) [57]. Volumetric capacity of C-MEMS sheets is shown in Fig 5-14. The fast ramp pyrolysis yields a carbon capable of inserting two times the amount of lithium per unit volume (≈400Ah/l) than the other samples. This result is in accord with the higher measured density of that sample (≈2.2). Density alone is not capable of explaining the factor of two increase of fast ramp pyrolysis. There must be a difference in the microstructural composition of that carbon compared to others. Although this was not evaluated in this study, the morphology of the resulting carbon is shown in Fig 4-1 . An increase in nanopore density was reported to have an increased level of reversible and irreversible lithium intercalation in pyrolyzed epoxy novolac carbon [58]. The increase in irreversible capacity for the fast ramped carbon is not apparent in Fig 5-15.
Irreversible Capacity
Irreversible capacity is defined as the difference between the capacity of the first intercalation and the reversible capacity of later cycles divided by the first intercalation capacity. The first intercalation is also when a solid electrolyte interphase layer (SEI) is formed on the surface of the carbon and possibly in the inner pores of some carbon [60]. 139
The irreversible capacity of pyrolyzed C-MEMS is quite elevated when compared to graphite (circa 12%) [61]. The different samples exhibited an average of 68% except for the forming gas treated sample which showed 85% of irreversible capacity. This high level of irreversible capacity is typical of hard carbons which can be close to 30% (P.100 [62]). Contamination from the electrolyte solution or the Viton o-ring could possibly play a role in increasing the irreversible capacity.
Capacity at Higher Rates
Figure 5-16 shows the performance of different C-MEMS carbon sheets under different discharge rates. The current densities can be easily obtained by dividing the net rates used by the area of the carbon exposed to the electrolyte (0.636 cm2). The drop in capacity over the range examined is accentuated by the testing architecture that was used. The 3mm of liquid electrolyte separator ensured a really high internal resistance of the system with Rsolution approximately equal to 40 Ω. For the purposes of comparison, between CMEMS layers, the setup is good enough since the Rsolution is the same for all the samples. The capacity is inversely proportional to the rate of discharge used. The corresponding values of volumetric capacity are shown for the 3 mA for the sake of comparison (see Fig. 5-17). In order to compare the relative performance of the different layers, the percent loss between the discharge at 0.1 mA and the discharge at 3 mA is evaluated and shown in Figure 5-18. For the standard conditions, the 12.5 µm film suffers the least percent loss (53%). The thicker 51 µm layer .loses 65% for the same change in discharge rates. The thinner layer performs better under high discharge rates regardless 140
whether the standard process or the fast ramp is used. Forming gas on the other hand seems to dramatically affect the performance the high rate performance. The percent loss for the forming gas sample was 72% which is 35% larger than the loss of a layer of almost the same thickness but pyrolyzed under N2. The percent loss seems inversely proportional to the estimated Li diffusion coefficients at 50mV vs.Li/Li+ (except for the 51um layer), as seen in Fig.5-25. Given that the intercalation process for long periods of time (over a few seconds) is governed mostly by material diffusion inside the solid carbon phase, a larger coefficient of diffusion would mean the anode can easier intercalate and deintercalate Li.
EIS Analysis
EIS was performed at 50mV vs. Li/Li+ at the end of each charge cycle after 5 minutes of potentiostatic equilibration time spent at 50mV in order to avoid the transients effects of the charging process. A peak to peak voltage of 10mV was used. The scans were done between 200 KHz and 100 mHz cycled twice. The typical outcome of an EIS is shown in Figure 5-19. For the sake of demonstration one of the samples from the 12.5 µm thick samples is presented. The Nyquist representation of the EIS results shows a system’s ohmic resistance of around 55 Ω corresponding to the high frequency impedances. A transient mode controlled by the charge transfer kinetic is a slightly depressed semi circle with a diameter of approximately 60Ω. The top of the quasi semi circle at approximately 4 KHz. At around 35 Hz. A diffusion controlled impedance as indicated by the 45o slope of the impedance curve follows at smaller frequencies. The impedance curve starts to 141
bend slightly downwards for even smaller frequencies were complicated mass transfer related effects control the impedance. Although visual inspection is capable of extracting to a reasonable approximation most of the parameters needed for analysis of a Randles cell model shown in Figure 5-20, this model is more fit for a situation where the transient part dominated by the charge transfer reaction is an exact semi circle, and the inquiry about the diffusion controlled area is is foregone. This is not the case for lithium intercalation in C-MEMS sheets where Semi circle is apparently slightly depressed and it is of our interest to investigate the diffusion inside the solid carbon phase. A modified Randles model containing a constant phase element and Warbug element is thus suggested (see Fig). In Figure 5-20, Rohmic represents the sum of electronic and ionic resistances in the system including the solution ionic resistance, the electronic resistance of the carbon and the current collectors, as well as the ionic resistance of any layer that forms at the surface of the electrodes. RCT refers to the charge transfer resistance of the lithium intercalation into the carbon. It is inversely proportional to the exchange current density which indicates the rate at which the intercalation and deintercalation occur. “W” refers to the Warburg admittance element in our case. This is expressed as:
W = Yo
jω
The Warburg element is used to represent the solid diffusion effects at low frequency regimes P.380 in [63]. Q is referred to as a constant phase element because it effectively
142
acts an element that shifts the phase independently of the frequency. When the shift angle 90o Q becomes equivalent to a capacitance, but at all other times it is different in nature.
Q = Yo ( jω )
n
Yo which can be equivalent to a capacitance when n=1 normally has units of S.sn (i.e. Siemens x seconds n) Q is sometimes related to the surface roughness of electrodes while some argue it can also be caused by a distribution of oxidation states at the surface of the electrode under study [64, 65]. Zsimpwin software is used to fit our CPE model to the impedance data measured. The curve fitting is accurate to below 2% error for all analyzed data included in the results. Figure 5-21 shows the Nyquist plots of the different C-MEMS carbon sheets. It can be seen that the Rohmic is the same within a 5Ω interval for all samples. The diameter of the semi circle indicates the RCT. It can easily be seen that the RCT is lowest for the fast ramp sample but for a more in depth analysis, Zsimpwin is used with an R(Q(RW)) which corresponds to the model of Fig. 5-20 (a). A summary of the extracted parameters is shown in table 5-3.The standard deviations on some of the values are quite large but a few trends emerge and will be discussed below.
Charge Transfer Resistance
The charge transfer resistance RCT for the different C-MEMS carbon sheets is shown in Fig. 5-22. 35 Ωcm2 is an approximation that fits most samples except for the 51 µm 143
pyrolyzed under N2 which has a value of 26 Ωcm2. Charge transfer resistance is highly dependent on the surface and although the starting materials are slightly different, it was expected that the Forming gas and fast ramp samples show substantially different RCT. Since most samples are in the same range, it can be inferred that the surface of the samples resulting from the pyrolysis of SU-8 is independent of the atmosphere used or the ramp rate. The 51 µm sample does deviate statistically much from the rest of the samples (because of overlapping standard deviations). The value reported in the literature for electrodes fabricated with 44um graphite particles is approximately 37ohmcm2 (approximate thickness of 150 μm, at 55mV)[66]. In general the reported values for Rct in the literature fluctuate between 5-35 ohm cm2 [67]. It is hard to make a direct comparison between C-MEMS electrodes and graphite electrodes which are composed of binder (e.g. PVDF) and packed graphite particles to an approximate thickness. What can be said is that the intercalation rate at the surface of C-MEMS electrodes is comparable to that of graphite electrodes in general.
Another way of looking at the charge transfer resistance is by finding the equivalent exchange current density. The relationship between the two is seen in Eq. (5.1): I0 =
RavT nFRCT
(5.1)
Where I0 is the exchange current, Rav is Avogadro’s constant, T is the absolute temperature, n is the number of charge, F is Faraday’s constant, RCT is the charge transfer resistance. Figure 5-23 shows the values of the exchange current density. 1.7mA/cm2 is the typical value representing the samples. Reported values in the literature indicate a 144
variation of the exchange current from 1.4 mA to 2.4 mA depending on the state of charge of the electrode [67](this value was for a PC:DMC mixture and for 100 μm graphite carbon sheet). The exchange current was smaller for higher SOC.
Double Layer Capacitance
The double layer capacitance information was very diffuse as seen in Table 5-3. a value of 2 µF/cm2 is indicative of the range of capacitances seen at the surface of C-MEMS. This value is quite low indicating the lack of charged groups at the surface of the carbon. The value of capacitances reported in the literature varies between 2 µF/cm2 for the basal plane of highly oriented pyrolytic graphite (HOPG) to 70 µF/cm2 for the edge plane of HOPG [68]. This is in conjunction with the literature that reports a very inert surface [56]. The double layer capacitance on the other hand increase between the first charge cycle and the third cycle. The percent increase varied dramatically between sample but the trend was clear. This can be associated with the deposition of a granular layer of SEI that slightly increased the surface area.
Lithium Diffusion in C-MEMS sheets
The diffusion of lithium inside carbon sheets is of great interest to us. It is an indicator of how the carbon will perform under high current loads. The solid state diffusion 145
coefficient of lithium inside carbon reported in the literature varies anywhere between 107
to 10-15 cm2/s [69] .In our case, the diffusion coefficient was calculated using the EIS
method followed from references [70, 71]. The theory assumes semi infinite plane diffusion. The state of charge at 50mV was assumed to be 90% for the purpose of the calculation. The charge curve at 0.1 mA hour was used an approximation of the titration curve. The values obtained showed a low diffusion coefficient approximately 3x1012
cm2/s. The values of the diffusion coefficients were related to the performance at higher
current. The higher diffusion coefficient of the 12.5 µm sample (5.5x 10-12 cm2/sec) was reflected with the smallest percent loss when the 0.1 mA and 3 mA discharge capacity were compared (see Figure. Similarly the sample treated with forming gas had the highest percentage loss from the high current effect and the smallest lithium diffusion coefficient (2.31 x 10-12 cm2/sec). The 28 µm sample falls in between the two with the same average performance at high rates. The 51 µm sample is inconclusive.
Lithiun vs Lithium
When typical electrochemical experiments are conducted, a strong effort is put towards making sure that the counter electrode has a quite larger area (circa 10 times) the area of the working electrode. In the case of out construction the area’s were very similar, so it was very important to make sure that the kinetics on the working electrodes were never limited by those on the counter electrode. An EIS was conducted by placing a Li foil instead of a regular carbon film. So the system really consisted of two lithium foils with a 0.636 cm2 separated by a 3mm electrolyte. The resulting Nyquist plot is seen in Fig.5-26, 146
along with a regular 12.5 μm C-MEMS film. The scaled charge transfer resistance of the Nyquist plot displayed is 14.7 Ωcm2. This value corresponds to two times the charge transfer resistance of a single sheet of lithium, since in the equivalent impedance circuit, both resistances are in parallel. Thus Rct of a single layer of Li film is 7.35 Ωcm2. This corresponds to a Li+ plating exchange current of approximately 8mA/ cm2, which is almost five times higher than the of the C-MEMS sheets. The 7.35 Ωcm2 can offset the value of Rct perceived from the EIS but it would not change the comparative nature of the different results.
5.1.4. Conclusion We have introduced a method for fabricating small carbon films from layers of SU-8. The layers can be lithographically defined and modified at will to optimize for parameters such Cdl or high power operation. The lithium intercalation properties of the given layers have been identified under different pyrolysis conditions and thicknesses. Specific capacity, volumetric capacity, lithium ion diffusion coefficient have been experimentally determined. A higher lithium diffusion coefficient ensured a smaller capacity degradation under higher discharge rates. Fast ramp pyrolysis under nitrogen (15 o
C/min) from room temperature without softbake was found to increase the volumetric
capacity of the carbon layer by two (making it 400 Ah/l). Thinner sheets pyrolyzed under nitrogen were the most resilient to high rate discharge. Forming gas environment during pyrolysis increased the irreversible capacity of the carbon layer. The intercalation depth of Li at intercalation rates of 160 μA/cm2 was shown to be at least 51μm.
147
5.2.
Enhanced FEA Modeling Approach for 3D
Microbatteries. 5.2.1. Introduction
The modeling of microbatteries is of interest to the microbattery fabrication technology. Ever since it has been identified that microbatteries can lead to enhanced capacity and power densities when compared to regular thin film microbatteries [72]. Several approached have been utilized to model batteries operating under different conditions. From the purely analytical modeling [73], to first order approximations, most of these techniques are aimed at capturing more or less some aspect of the operation of a microbattery. The aim of our approach is to focus on the electronic and ionic current densities inside a battery which are important parameters that need to be taken into consideration when choosing particular microbattery architecture. In the arena of 3D microbattery modeling, Hart et al have proposed that a method relying purely on a conductive media model is capable of capturing the current density modeling in a satisfactory manner [74]. It is argued from the bases of experimentally performed Electrochemical Impedance Spectroscopy EIS that the method of white et al. does not capture the correct current density distribution picture because it completely neglects the dominant component in such a simulation the Charge transfer resistance. The current 148
work proposes an adjustment to the White paper that allows it to better capture the image of current densities in a technique that does not add much complexity to the FEA analysis.
5.2.2. Method Specifics and Techniques
In a typical Randles cell representation of a simplified electrochemical system, the notable components are the Rohmic and Rct. While the Randles model does not capture the mass transfer limited slow kinetics it’s a good approximation of the internal resistance of a battery at a given state of charge. The Randles cell is portrayed in Figure 5-27. The Rohmic corresponds to the sum of the electronic resistance of the electrodes and current collectors as well as the ionic resistance of the electrolyte solution/separator layer that occupy the space in between the electrodes. The Rct is specific to the chemical reaction that is occurring at the surface of the electrode and more specifically at the surface of the anode in our case.
149
Figure 5-27 Randles cell representation of a simplified electrochemical reaction. The Nyquist plot of such a model is a semi circular model shown in Figure for specific values of Rohmic 5ohm x cm2 and Rct 20 ohm x cm2 and Cdl 3 μF.
-Im(Z) vs. Re(Z) 10 9 8
-Im(Z)/Ohm
7 6 5 4 3 2 1 0
5
10
15 Re(Z)/Ohm
20
Figure 5-28: Typical Nyquist plot of a Randles cell 150
25
The impedance of the cell at higher frequency is reduced to the ohmic component. While it’s impedance at low frequencies is a the sum of Rohmic and Rct. White et al. chose to represent the system using uniquely the Rohmic to capture the image of the current densities for a specific electrode construction. While some battery systems with highly resistive electrodes and very facile charge transfer can be represented in this manner, it is not the case for a lot of different battery systems where Rct is the dominant contributor to the impedance at steady state (e.g. Li ion insertion based batteries). This Rct dominant factor is specifically located at the surface of the electrode (if the electrode is non porous, which is expected of very thin 3D microbattery construction). Our work includes this charge transfer resistance in a relatively thin layer of highly resistive material. Doing this allows us to capture the important effect that the rate of the chemical reaction imposes on the current density profile at quasi steady state, while keeping a purely conductive medium FEA solving technique. Since the charge transfer resistance is accounted for in this model, it is estimated that the results are closer to the real image of the current distribution in a battery.
Method and Techniques
The technique is tested out for a simplified case where the “battery” architecture uses a liquid electrolyte as a separator between a flat cathode and a microstructured carbon anode. In effect, this model is tested out on a half cell first where the one of the electrodes
151
is a flat lithium ion film and the other is layer of microtextured carbon that can be fabricated using the techniques described in section 2.1. Model description
The model considered mimics the design of a cell used to test microstructured carbon half cells. The focus is on area of approximately 700 μm width containing carbon microposts 100 μm high with a similar and 50 μm wide. The conductivity of the materials is set from know values in the literature to mimic pyrolysed SU-8 at 900oC (1x104 S/m)[27] and electrolyte conductivity properties of EC:DMC 1M LiClO4 (≈0.8 S/m) [75, 76]. The resistance equivalent to the charge transfer resistance will be determined with the help of experimental values. The layer that represents the charge transfer is a 2 μm thick layer covering the carbon anode. This layer does not have a real physical equivalent to it since the typical charge transfer occurs across an SEI layer that can be as small as a few nanometers or could grow to a micron or so. The importance in that layer is that it represents a conservative geometric upper bound beyond which the potential would be very similar to what is in reality while taking into account the charge transfer (regardless of the thickness of the SEI layer). The value of the resistance attributed to that layer in the modeling is determined in a way to fit the potential distribution that is harmonious with EIS results obtained through experimental testing
152
Experimental Values Extraction
The details of the EIS testing have been previously presented in section 5.1.2. In short and EIS with 10mV peak to peak perturbation was done at 50mV vs Li/Li+ on a 51 μm base layer. The extracted value from the Nyquist plot is shown in Fig. 5-29. The Rohmic was estimated at 52 ohms and the value for Rct was estimated to be 42 ohms. The values are close enough to dictate that half of the potential drop of potential in the electrolyte (majority of Rohmic) should be half that of the total potential drop (3V applied in our case).
EIS Nyquist Plots 20
-Im(Z)/Ohm
15 10 5 0 -5 -10 40
60
80 Re(Z)/Ohm
100
Figure 5-29 Nyquist plot of an actual carbon layer in an EC:DMC 1M LiClO4 electrolyte at 50mV Vs Li/Li+.
153
A trial and error method for determining the value of resistance that should be attributed to the 2 μm layer is performed. The process is not very tedious since the simplicity of the conductive medium two dimensional model chosen allows quick convergence (less than 15 sec) even for slower machines. The result shows that close to the electrode the potential is approximately at 1.5 V for a conductivity of the layer equivalent to 2.5x10-4 S/m. It is obvious from the potential distribution observed in Fig 5-30 that a large potential drop occurs across the charge transfer resistance equivalent layer, which highlights its importance in the overall current density distribution picture.
Figure 5-30: Snapshot of the modeling of the electric potential distribution when a 2.5x10-4 S/m conductivity is attributed to the thin 2 μm charge transfer equivalent layer. The electrolyte layer extends way beyond the top to reach 3 millimeters (3x10-3 m) which is the actual separation between the working and reference electrodes in the real cell.
154
Results
Figure 5-31: Norm of the current density distribution in the modeled area.
155
Figure 5-32: Current density profile at the top of the electrodes (10 μm below the top level of the electrodes).
156
Figure 5-33: Current density at the bottom of the electrodes (10 μm above the base of the electrodes).
Current distribution results
It appears from the plots of Fig 5-31 through 5-33 that the current distribution in the high aspect ratio carbon post goes from being 1/3 the current density in the electrolyte at the top of the posts and becomes three times the value of the current density inside the electrolyte towards the bottom of the post. The distribution is expected when one considers that the voltage drop across the very thin 2 μm charge transfer equivalent layer dominates over the geometric configuration of the electrodes. In a sense, the Delta R imposed by the geometry is minimal compared to that imposed by the Rct which 157
commands a quasi uniform current density at the surface of the Rct layer. A uniform current density at the surface of the charge transfer layer imposes that electrolyte carries all the load required to satisfy that condition. Looking at it from purely geometrical perspective, the ratio of average current densities between the top and the bottom of the electrolyte channel should be equivalent to 50 μm (section of 1 channel)/250 μm (section of having Rct ) factor of 5. Although a factor of approximately 3 is observed in the simulations this analysis captures the qualitative description of what is occurring. The discrepancy can be caused by the high sharp curvature of the elements at the corner of the electrode leading to the distortion of the results through a numerically induced error. It is also very interesting that the results remain almost unchanged when the electrolyte is replaced with a highly conductive material (the same value as carbon) which indicates that the current distribution seen above is also valid for the case of interdigitated anode cathode battery architecture and that similar current density concentration profiles are to be expected.
Comparison with Long and Hart et al.
A way to look at the effect of the Rct addition to the simple conductive medium model proposed by Long and Hart et al. we compare the potential distribution in the two model on identically constructed electrode architecture with triangular elements in an alternating cathode anode configuration as seen in Figure 5-34 taken from [72].
158
Figure 5-34: Isopotential lines shown for a triangular 3D microbattery construct with alternating cathode and anode from Ref.[72].
Figure 5-35: Electric potential distribution including the Rct effect for the same architecture presented by ref [72] shown in Fig.5-34
159
Figure 5-36: Current density distribution map of the triangular architecture model
The triangular electrodes constructed to mimic the model of ref.[72] were modeled with a 45 μm side and a 25 μm vertical separation between electrodes. By comparing Figure 534 and 5-35, it can be seen that the potential drop in our model is occurring abruptly at the surface of the electrode in contrast with the [72] model that exhibits a distribution of isopotential line in between the electrodes. This is validated by the very uniform current density distribution in the electrolyte space (see Fig 5-36). This demonstrates how the inclusion of Rct can totally change the image of the current distribution in a 3D microbattery model and in general a smoothing effect on the current density would occur. This alleviates the concerns that Long and Hart et al. have put forward in with regards to
160
non uniform current distribution between microelectrodes used in the context of 3D microbatteries [72, 74].
Validity of the Model
From the mass transport equations of ionic species in solutions, the generalized NernstPlanck equation applies for the total current density through any point inside the solution (P. 138 in [63]):
J j = − D j ∇C j −
zjF RT
D j C j ∇φ + C j V
(5-1)
Where J is the current density of a specific species j, D is the diffusion coefficient of that species in solution, Cj is the concentration of species j, F is Faraday’s constant, R is the gas constant, T is the absolute temperature, zj is the number of charge on species j, V is the velocity of the bulk fluid at the specified location.
In the absence of convection, which is usual for small unstirred solutions that are not strongly heated, the convection term is neglected. The first and the second on the right hand side of the Nernst-Planck equation represent the diffusion and the migration transport respectively. When evaluating that equation in the bulk of a solution, the diffusion transport is often neglected because concentration 161
gradients are considered to be too small. The contribution of migration effects on electroactive species are considered negligible when highly concentrated non electroactive species are present. These species (often a strong electrolyte) work very well as ionic conductors and minimize the total resistance of the system and reducing the effects at the electrode of electroactive species to a diffusion effect and therefore simplifying the decoupling between migration and diffusion. On the other hand, very close to an electrode, both migration and diffusion effects have a non negligible contribution to the current density and care should be used in making generalization from simplified models. The 3D microbattery configuration is an example where such care needs to be taken. The results can be right or wrong and strongly dependent on the exact composition of the electrolyte. If the thickness of the diffusion layer is large enough to encompass the neighboring electrode, a conductive medium modeling approach becomes inappropriate for the analysis. The thickness of the diffusion layer is strongly dependent on the concentration of the electrolyte, the electrolyte species, and the reaction rate at the surface of the electrode. In the case where the diffusion layer extends from one electrode to the other, the concentration gradients inside the solution become substantial and can easily affect the flux of ions under study and the conductive medium modeling approach becomes incapable of predicting the right current density profiles. In fact, if you use a value of the diffusion coefficient of lithium salts in an EC:DMC close to 1x10-5cm2/s [77] and under the assumption that a limiting current occurs at almost 1 mA/cm2 .That assumption leads to a lower limit estimation for the thickness of the boundary . From the limiting current equation for that case: 162
il = nFA
Do
δo
C o*
(5-2)
Assuming il is 1 mA/cm2, n being one for Li+ , F is 96500 , A is 1cm2, D is 1x10-5cm2/s and Co is 10-3 mol/cm3 or 1M. δo the steady state diffusion layer thickness turns out to be approximately 10microns. Since assuming a limiting current condition leads to a lower limit on the diffusion thickness layer, this 10 μm is a conservative lower bound of a where the influence of diffusion cannot be neglected. It is safe to say that in cases where the model is looking at electrodes with a spacing smaller than 10 microns, both this model and any approximation used in [72, 74] would be conveying a distorted image of what the real current flux distribution in an electrolyte is.
Our model assumes that all charges responsible for electronic conduction can lead to a electrochemical event a the surface of the electrode. This assumes that the transference of lithium in the solution is close to 1 which is not the case because part of the conductivity is due to the ClO4- ion. Although the error resulting from this assumption would lead to an offset of the results, but the general qualitative assessment still remains valid. This general approximation is good enough for a qualitative understanding of the current density distribution. The results when compared to the Long and Hart et al. case should yield a smoother current distribution.
163
5.2.3. Conclusion
An enhanced FEA model of 3D microbatteries has been proposed. A combination of modeling parameters and experimental data was utilized. A 2 μm thin layer that represents the chemical reaction’s effect on the on the current density distribution has been proposed, for the first time, as a useful tool to capture a more exact picture of the current densities in a microbattery.
164
Chapter 6
6. Other Applications of Carbon MEMS in Electrochemistry
6.1.
Dielectrophotretic
Trapping
for
Novel
Photolithography Techniques
6.1.1. Introduction
The electronics and photonics industries are very interested in developing new nanolithography methods in order to continue the long-term trend of building ever smaller, faster, and less expensive devices (Moore’s Law!!). Conventional projection lithography is based on the 4x reduction of reticle features and is limited by optical diffraction where the smallest feature size achievable is approximately the order of the 165
wavelength of the radiation source [20]. For wavelengths shorter than 157 nm, refractive optics are no longer suitable because of the strong absorption of shorter wavelengths by the optics and much more expensive and demanding reflective optics must be introduced.
There is no consensus yet on what the next generation lithography process will be. Instead, there are many technologies that are competing. Extreme Ultra-Violet (EUV) lithography, X-ray lithography, e-beam and ion-beam lithography, soft lithography, and proximal probe lithography are all potential contenders in this crowded arena.
The most logical extension of the lithography used today is to use shorter wavelengths. Although progress has been made by using smaller and smaller wavelengths, in Extreme Ultra-Violet (EUV) this approach has many inherent challenges [78]. In the latter case, in order to use 13.4 nm EUV light for device fabrication, an all-reflective optical system under vacuum is required.
The cost and reliability of these systems are major
disadvantages of this approach.
Resolution Enhancement Technologies (RET) such as Phase Shift Masks (PSM), OffAxis Illumination (OAI), and Optical Proximity Correction (OPC), liquid immersion lithography [79] and interferometric lithography [80], all enable the writing of photoresist features with dimensions smaller than the wavelength of the exposing light source. The problem with RET is that the added manpower required, the extra computing and databases, and software and technology licensing is very expensive. 166
Using an e-beam as a “light” source for projection lithography has also been developed. The method, called SCALPEL [81, 82][9-13] has many advantages over conventional lithography including smaller features and better depth-of-focus, but the price of the mask fabrication and equipment is a major disadvantage.
Soft lithography [83] is another methodology that tries to achieve small dimensions in a parallel manner.
Since soft materials are used, deformation of the stamp/mold,
reproducibility (due to distortion), and defects (yield) are problems that need to be resolved [83]. Nanoimprint Lithography and Step-and-Flash Imprint Lithography are also schemes where a mold is used to create nanofeatures, but these schemes have problems in defects and scalability.
With AFM [84, 85], STM [86],[87], NSOM [88],[89] ANSOM [90], electron-beam lithography, “Dip-pen” nanolithography [91] or other sequential writing schemes, smaller features can be defined, but these technologies are expensive to implement and because of their serial nature they are also too slow. Arrays of these have been researched, but in the case of the proximal probe approach, the tip stability is an issue that needs to be resolved as demonstrated by the work on IBM’s AFM arrays (the so-called millipede [92],[93]). One can also write photoresist features with dimensions smaller than the wavelength of the exposing light source by using evanescent waves, also known as nearfield light, as was demonstrated recently with Near-field Scanning Optical Microscopy
167
(NSOM) [88, 89] and Aperture-less Near-field Scanning Optical Microscopy (ANSOM)) [90].
Plasmon printing, also demonstrated recently, is a variation on the evanescent wave approach described above [94]. In this method, surface evanescent waves resonate in illuminated silver nanoparticles and dramatically enhance the electromagnetic field near the particle’s surface. It was found that the silver nanoparticles actually resonate at a frequency that sensitizes g-line photoresists. Using a conventional g-line photoresist and a conventional g-line UV source, features 30~60 nm in width and 10~15 nm in depth were obtained [94]. This result demonstrates that using evanescent waves, features approximately one-tenth the wavelength of the UV light source can be written.
Nanopatterning of surfaces by near-field enhanced laser irradiation has also been demonstrated [95-98] . In this approach particles are used to focus pulsed laser light to literally burn patterns into the substrate.
In this chapter, the feasibility of a system where local environment changes around a nanoparticle suspended in a liquid are used to pattern sub-micron features on a sensitive surface is investigated and several example systems are proposed.
Although the
nanoparticle-induced changes can be chemical, electrochemical or optical, in this chapter, only optical methods will be discussed. Likewise, although different methods can be used to actuate (move) the particle, in the current contribution, only electrokinetic means 168
will be described. With such a system, diffraction limitations can be overcome by making use of near-field light to pattern nanoscale features. Multiple particles could be driven at the same time to allow parallelism in fabrication.
In working with nanoparticles and nanofeatures, researchers have come to realize that, for further progress in nano-manufacturing, one might have to work in liquids rather than in air. Very small quantities of materials have huge surface to volume ratios, leading to fast evaporation, easy contamination, difficulties in manipulation of particles, etc. Nature has always made things from the bottom up, working with very small particles. Its nucleic acids and proteins are kept in a liquid environment within a cell; call it a water management system for its smallest building blocks. Along this same line of reasoning, several years ago, IBM’s Romankiw spelled out the many advantages of working in liquids for the fabrication of even today’s ICs and MEMS, pointing out, for example, the much higher fidelity of electrochemical deposition compared to any gas phase deposition process [99]. Here, it is investigated whether this thinking extends to nano-lithography.
6.1.2. Particle-Based Nanolithography
We define particle-based nanolithography as a method of creating submicron traces on a sensitive substrate by using a particle (or particles) as a transducer element. The particle emits, channels, or reflects energy towards the sensitive material during the writing process. In figure 6-1, the basic concept of particle-based nanolithography is illustrated. 169
Transport Field Particle Trace
Sensitive Material
Particle
Energy
Substrate
Figure 6-1: Illustration of the basic concept of particle-based nanolithography.
In order to validate the feasibility of such a system, the various system requirements must first be addressed. In this chapter, the feasibility of a particle-based nanolithography scheme using light as the sensitizing energy and electrokinetic means as the actuating force is investigated. In order for the system to be feasible: First, the intensity of the light emanating from the particle must be strong enough and the substrate must be sensitive enough for a trace to be defined. Second, the actuation method must be reasonably accurate and precise. (For example, Brownian motion will introduce error in the motion of very small particles.)
Finally, the process must be fast enough for significant
throughput or be easily parallelizable. We used relatively large microparticles (2.8 µm – 10 µm) for validation studies and theoretically examined the effects of scaling down to the sub-micron range. 6.1.3. Particle – Substrate Compatibitlity
170
The challenge of finding a sufficiently sensitive substrate for a moving nanoparticle to write on (i.e., a particle that induces enough photoevents to sensitize the substrate) can be illustrated by an example based on a two-photon fluorescent particle and a conventional g-line photoresist. We are using a two-photon particle here to avoid direct exposure of the resist by the illuminating source; in a two-photon absorption process the emitted light has a higher frequency than the incoming radiation so only light emitted from the particle will expose the substrate. According to our calculations, for a 30 nm diameter particle moving at 100 nm/s to sensitize a thin photoresist layer with an exposure dose of 50mJ/cm2, an illumination intensity of 7.348 Gw/cm2 at 700nm wavelength is required. This is an intensity that can be realized (the record intensity is 1018 W/cm2), but it is very difficult with current technology to achieve such high intensities (high cost, short durations, extreme focusing). Typical laser intensities are much lower, especially for continuous lasers.
Thus, an amplification scheme is needed.
One example of an
amplification scheme is the development of silver halide grains in traditional photography. In silver halides, one photon leads to 108 silver atoms, an amplification of 108! The exposure dose of conventional photoresist in g-line exposure systems is on the order of 100 mJ/cm2. (For example, Shipley 1827 photoresist has an exposure dose of 150 mJ/cm2.) Photographic film requires much smaller exposure doses. For Kodak TMAX p3200 professional film (which is inexpensive and readily available at any pro photo store), the exposure dose for significant contrast calculated from published Spectral-Sensitivity Curves is on the order of 1 nJ/cm2. This is a difference of 8 orders of magnitude in required exposure dose! 171
Another way of preventing exposure of the substrate from a flood exposure source is to use phosphorescent particles. Phosphorescence is a phenomenon where particles store energy and exponentially release it over a long period of time. The particles themselves would be the light source in this case, and they would be illuminated in a zone away from the photo sensitive substrate.
We have demonstrated that regular film can be used as a rapid and inexpensive tool for detecting the movement and particle location of weak phosphorescent particles. Strontium Aluminate-based violet-light-emitting phosphorescent particles (Shannon Luminous Materials, CA) were used in our experiments.
After exposing the
phosphorescent particles to g-line UV light for 10 seconds at 5 mJ/cm2, various photoresists were exposed to the particles for extended periods of time (30 minutes and longer).
None of the photoresists developed any noticeable patterns.
Exposure
experiments were also done on Kodak TMAX p3200 professional film with particles that were similarly charged for 10 seconds at 5 mJ/cm2. Exposure patterns created by single particles were easily observed even for 30 second exposure times as demonstrated in figure 6-2. The particle diameter and the pattern diameter were comparable in size (30 – 50 µm). Particles were also dragged across the surface of the film in a step-wise manner for a total time of 10 seconds and a total distance of 1.5 mm. A trace consisting of a sequence of dots connected by less defined lines developed, as shown in figure 6-3. A picture of unexposed film that was developed under the same conditions is shown in 172
figure 6-4 and was used for control purposes. The average grain diameter for typical high-speed roll film is approximately 1 micron in size. The resolution of the film is also of the same order of magnitude.
Figure 6-2: Exposure patterns from single phosphorescent particles (approx. 40 μm in diameter) on Kodak TMAX p3200 professional film. (The exposure time was 30 seconds.)
(Figure 6-3.a)
(Figure 6-3.b)
Figure 6-3.a Exposure pattern from dragged phosphorescent particles (approx. 1.5 mm in length). 173
Figure 6-3.b Close-up of the line trace seen in figure 6-3.a
Figure 6-4: Control picture of Kodak TMAX p3200 professional film only exposed to ambient light in the dark room for 30 seconds (Developed grains are visible due to the ambient light in the dark room.)
6.1.4. Nanomanipulation Issues
There are many methods available for actuating (moving) a particle in a liquid. Electroosmosis, electrophoresis, dielectrophoresis, optical tweezers (optical trapping), magnetic manipulation, pressure-based movement, electrostatic manipulation, Local area Melting Micro-Manipulation (LMMM) [100], heat convective flows, and acoustic actuation are all techniques that allow movement of a particle suspended in a liquid 174
without actually touching the particle. Of these techniques, the electrokinetic methods (electroosmosis, electrophoresis, dielectrophoresis) tend to scale most favorably when scaled down and are easily controlled by simple application of voltages on surface electrodes. Also, electroosmosis in particular has the advantage of moving a thin layer of liquid very close to the surface. Since electroosmosis is a bulk fluid phenomenon instead of a single particle actuation technique many particles can be actuated in the same manner at the same time allowing parallelism. We will address possible scaling issues in electrokinetic manipulation in Section 6.1.5.
6.1.5. Scaling Laws
SCALING OF ENERGY EMISSION
For volumetric energy emissions where the energy is emitted from or channeled through the bulk of the particle (e.g. phosphorescence and fluorescence), the energy emitted from 3
the particle scales with the volume ( r ). If a particle is coated with a luminescent 2
material, the energy emission scales with the surface area ( r ). As the particles are scaled down to the submicron range, the energy emitted and the energy that can be stored within the particle are both greatly diminished. A particle that acts as a transducer for an 175
external energy source will be able to deliver higher exposure doses compared to a particle that stores and releases energy, especially over long exposure times.
Nanomanipulation Scaling Scaling of dielectrophoresis
In dielectrophoresis, the difference of polarizability between particle and solution in a non-uniform electric field gives rise to a net force acting on the particle. In positive dielectrophoresis, particles are more polarizable than the solution and tend to move toward high-field regions, and in negative dielectrophoresis, particles are less polarizable than the solution and migrate toward low electrical field regions. (See chapter 3.) The equation governing the magnitude of the dielectrophoretic force for a sphere is given as [101]:
⎡ ε~p − ε~m ⎤ 2 < FDEP / SPHERE >= πε m r 3 Re ⎢ ~ ~ ⎥∇ E 2 ε ε + m⎦ ⎥ ⎣⎢ p
ε m : the permittivity of the medium ε~p
: the complex permittivity of the particle
ε~m : the complex permittivity of the medium r E
: radius of the spherical particle : electric field 176
(4-1)
3
Thus, dielectrophoresis is a volume phenomenon that scales with the volume ( r ) of the particle.
The electric field between the energizing electrodes can be reasonably approximated as being inversely proportional to the distance between the electrodes at a constant voltage. If lc is defined as the characteristic length of the electrodes,
E
2
1 2 scales as lc .
It follows from Equation (1) that
∇E
2
, and thus the
3
dielectrophoretic force, is inversely proportional to lc . When scaling down the size of electrodes in a dielectrophoretic system, the dielectrophoretic force increases three orders of magnitude for every order of magnitude the characteristic length of the electrodes ( lc ) is decreased [101].
Conversely, the dielectrophoretic force
decreases three orders of magnitude for every order of magnitude the diameter of the particle ( r ) is decreased. If both the electrodes and particles are simultaneously scaled down, the scaling effects on the dielectrophoretic force tend to cancel out.
Scaling of electrophoresis
In electrophoresis, a Coulomb force is exerted on charged particles in a solution causing the particles to move with respect to the solution. The Coulomb force is given as: 177
r r FEP = QE
(6-2)
2
This force, felt by the charged particle, scales with the surface area ( r ) or the 3 volume ( r ) depending on the charge distribution throughout the particle. In most
cases, charges are distributed on the surface of a particle and electrophoresis is 2
usually a surface phenomenon that scales as r .
Since the electric field strength is inversely proportional to the distance between electrodes, the electrophoretic force is inversely proportional to lc , the characteristic length of the electrodes. For every order of magnitude the particle diameter is scaled down (for surface distributed charges), the electrophoretic force exerted on the particle is reduced by two orders of magnitude. When the electrode distance is miniaturized by one order of magnitude, the electrophoretic force the particle feels is increased by one order of magnitude. If both the particle diameter and the electrode size are simultaneously reduced, the electrophoretic force on the particle diminishes proportionally.
Scaling of electroosmosis
The equations governing electroosmosis and electrophoresis are mathematically identical (Coulomb’s Law), but electrophoresis describes the phenomenon where a 178
charged particle is moving in a liquid (see above), whereas electroosmosis describes the motion of a liquid on a charged surface. The Coulomb force acting on an infinitesimal volume element in the double layer is given as [101]:
r r FEOF = EρAdy
(6-3)
ρ : charge density of the medium in the double layer Ady :
infinitesimal volume element in the double layer
The electroosmotic force on the medium is thus proportional to the applied electric field and scales inversely with the electrode distance. The fluid flow results in a drag on the particles present in the solution. The fluid velocity at the slip plane is given as [101]:
r εζ r u = −E
η
(6-4)
r u : fluid velocity
ε : the permittivity of the medium ζ : the zeta potential in the double layer
η : the viscosity of the medium
179
The drag force (Stokes’ Force) acting on individual spherical particles at relatively slow fluid velocities is given as:
r r Fdrag = 6πηru
(6-5)
Thus, the Stokes’ force acting on individual particles is proportional to the size of the particles and inversely proportional to the electrode distance.
If both the
particle size and the electrode characteristic length are miniaturized at the same rate, the scaling laws of the particle and the electrodes tend to cancel each other out and the Stokes’ force on the particle due to electroosmosis remains the same.
Summary of scaling in dielectrophoresis, electrophoresis, and electroosmosis
From the previous sections (4.5.2.1 – 4.5.2.3), when simultaneously scaling down particle size and electrode size, the use of dielectrophoresis and electroosmosis is advantageous compared to the use of electrophoresis. The effects of particle and electrode scaling cancel each other out in the cases of dielectrophoresis and electroosmosis, but in the case of electrophoresis the weakening of the force due to 2
3
particle miniaturization (scales as r ~ r ) dominates over the strengthening of the force due to electrode miniaturization (scales with the inverse of r ).
Brownian motion
180
Some
phenomena
that
could
potentially
interfere
with
particle-based
nanolithography become more apparent when the size of the moving particle is reduced. For example, Brownian motion becomes a critical factor when using submicron particles. The root mean square (rms) displacement (standard deviation) of the particle due to Brownian motion after time, t , is given in equation (6). From this equation, after a time t , a particle has a 66.7% probability of being found within the distance, σ
[101].
For a particle-based lithography system, the rms
displacement reflects the extent of the line broadening due solely to Brownian motion.
σ = 2 Dt
(6-6)
σ : rms displacement
D : diffusion coefficient of the particle t : time
The diffusion coefficient of a particle is defined as [101]
D=
kT 6πηr
(6-7)
kT: Boltzmann Temperature 181
As an example, the diffusion coefficient for a sphere of 1 µm diameter in water at room temperature (20˚C) is 4.28 x 10-9 cm2s-1. The standard deviation for this 1 µm sphere for a 1 second experiment is 0.926 µm. Obviously, diffusion effects will be critical for nanopatterning applications.
From Equation (6-7), it would seem that increasing viscosity reduces the error due to Brownian motion, but closer examination of the governing equations proves otherwise.
The velocity of a particle traveling in viscous medium can be derived from Stokes’ Law as:
r u particle =
r F 6πηr
(6-8)
For a lithography system to pattern a trace with length L, the time taken to complete the trace can be found as:
L L6πηr = tL = r r u particle F
(6-9)
t L : time required to trace a path of length, L
L: trace length 182
The rms displacement of a particle traveling a distance of L may thus be calculated by substituting Equation (6-7) and (6-9) into Equation (6-6) or:
σ L = 2 Dt = 2
kT L6πηr r = 6πηr F
2kTL r F
(6-10)
σ L : rms displacement of a particle traveling a distance L
In Equation (6-10), the viscosity terms cancel each other out. We conclude, rather counter intuitively, that the error introduced by Brownian motion (diffusion effects) is independent of fluid viscosity.
The diffusion error can only be minimized by operating at a lower temperature, patterning smaller electrodes, or by increasing the actuating force on the particle and thus increasing the particle speed.
Other scaling limitations
Although increasing the electric field magnitude and thus increasing the electrokinetic force on the moving particle by miniaturizing the electrodes can enable scaling down of the particle size to a certain point, it is naïve to assume that the particle can endlessly be scaled down. Other than the scaling laws detailed 183
above, effects such as Joule-heating-induced electrothermal flow, buoyancy, lightelectrothermal effects (if there is a strong light source present), and electrolysis of the solution all work to prevent scaling beyond a certain point [102].
6.1.6. C-MEMS
When using planar electrodes in dielectrophoresis, the electric field is confined to a volume that is very close to the surface. Taking advantage of high-aspect-ratio electrodes allows for better field penetration into the medium and stronger electrokinetic forces can be achieved at the same potentials. The use of 3D electrodes can thus be advantageous when miniaturizing a dielectrophoretic system. Two orders of magnitude improvement in trapping force for dielectrophoretic traps was observed for high-aspect-ratio gold electrodes [103]. 3D high-aspect-ratio carbon structures were used for the experiments to make use of this 3D enhancement effect.
C-MEMS electrodes can be easily patterned into complex 3D geometries that were previously difficult or expensive to fabricate using conventional carbon electrode fabrication methods. Also, the electrochemical properties of carbon make it an excellent electrode material [1]. For example, C-MEMS electrodes allow for a larger range of electric fields before the onset of hydrolysis (with its disruptive bubble formation due to the breakdown of water into hydrogen and oxygen). Furthermore, direct comparison of 184
survivability of carbon electrodes with gold electrodes in DI water reveals that carbon survives at higher potentials and that the failure of the electrodes is less catastrophic. In the proposed lithography systems here, there is a need to keep the particles as close as possible to the photosensitive surface in order for the trace from the particle to be better defined because of the need for near-field exposure of the substrate.
This can be
achieved using 3D dielectrophoretic traps. The 3D dielectrophoretic traps fabricated in C-MEMS trap the particles above the electrodes, enabling the trapped particles to be brought very close to the sensitive substrate (Figure 6-6). Also, the higher electric fields accessible through the use of C-MEMS electrodes enable stronger trapping. In fact, to the best of our knowledge, the Madou Research Group has fabricated the highest-aspectratio carbon electrodes ever created (Figure 6-5).
185
Figure 6-5: High-aspect-ratio C-MEMS structures. (Courtesy of Chunlei Wang, UC Irvine)
6.1.7. Proposed Lithography System and Validation
Two novel particle-based lithography systems are proposed and feasibility experiments are detailed.
From the previous analysis, it can be concluded that if the electrode
characteristic lengths and the particle sizes are scaled down together, the effects cancel out for nanomanipulation in the case of dielectrophoresis and electroosmosis. If other 186
effects such as Joule heating and hydrolysis are ignored, the scaling down of particlebased lithography systems is limited only by the particle-substrate sensitization threshold. Experiments in the microscale were performed to demonstrate the feasibility of the two example lithography systems.
Particle-Based Lithography System Using Dielectrophoretic Traps
Even though AFM lithography systems can easily define submicron patterns, AFM-based parallel writing schemes face considerable technical problems. The height of the AFM tips and the surface smoothness are hard to control and tip and medium wear are a problem that is inherent to AFM array contact methods, and consequently, system reliability is often compromised [92, 93].
The proposed system is very similar in
configuration to an AFM array, but since there is no direct structural connection between the actuating stage and the transducer element, the wear on the writing element is avoided. Particle-based lithography using dielectrophoretic traps is a scheme in which electrodes, preferably small carbon electrodes, are patterned on a surface in a pattern that can be used to trap particles at certain locations. As was discussed previously in section 6.1.6, the particle must be kept in the immediate vicinity of the photosensitive substrate for a fine trace to be developed. C-MEMS electrodes are excellent for use in the creation of 3D geometries, and C-MEMS electrodes have been used in fabricating 3D dielectrophoretic traps that levitate the particle above the electrodes. A dielectrophoretic trap uses negative 187
dielectrophoresis to confine particles in a potential well. Particles are held in the traps and then used in a manner similar to AFM arrays to create traces on a substrate. The stage that holds the dielectrophoretic trap array is moved in the x-y directions to create patterns on the photosensitive substrate. Because many identical traces are created all at the same time by controlling the stage, this method might result in a high throughput. The potential wells created by the dielectrophoretic traps hold the particles in place in the x-y directions, but let them move with some freedom in the z direction allowing a greater degree of leniency in the z alignment of the stage compared to AFM arrays. This scheme is illustrated in figure 6-6.
Figure 6-6: Illustration of particle-based lithography using dielectrophoretic traps. 188
To address the feasibility of creating a large array of 3D dielectrophoretic traps, an array of dielectrophoretic traps was fabricated using 3D C-MEMS posts for electrodes (Figures 6-7 and 6-8).
Pyrolysed photoresist was also used to create all the electrode
interconnects. The electrode height is 66 µm and the height of the interconnects is 9 µm. The posts have a center to center distance of 150 µm. The posts diameter is 35 µm, and the interconnect width is 26 µm. The experiments were done in DI water. DI water has the advantage of allowing high field strengths without Joule heating and hydrolysis due to the low conductivity of the medium. A 16 V 300kHz AC voltage was applied to the interconnects. Trapping of 9.62 µm polystyrene beads in these 3D traps is shown in Figure 6-9. The particles are suspended in a trap above the electrodes.
189
Figure 6-7: SEM of a dielectrophoretic
Figure 6-8: Higher magnification view
trap array fabricated from pyrolysed
of a single electrode post fabricated
Figure 6-9: A cluster of 9.62 µm polystyrene beads trapped in a 3D dielectrophoretic trap between 4 high-aspect-ratio posts. The particles are suspended above the posts.
More planar carbon electrodes can also be used as dielectrophoretic traps for use in particle-based lithography systems. In figure 6-10, trapping of 9.62 µm polystyrene beads in a negative dielectrophoretic trap composed of the more planar C-MEMS 190
electrodes with an electrode gap of 50 µm is shown. The carbon electrodes are at a height of 9 µm and the solution used is DI water. The same AC voltage as used in the previous experiments was used.
Figure 6-10: Illustration of a cluster of 9.62 µm polystyrene beads inside a dielectrophoretic trap made from flat carbon electrodes.
Parallel Electroosmotic Writing
Another scheme for writing with particles on a substrate in a liquid consists of using electroosmotic flow to move multiple particles in the x-y plane to trace patterns. We will refer to this lithography scheme as parallel electroosmotic writing. With the simple electrode configuration shown in figure 6-11, control of particles in the x-y plane was achieved.
When potentials are applied to opposite electrodes, motion of particles
between the electrodes is observed. Motion in the x direction, motion in the y direction, and motion in the diagonal directions were induced by application of the proper potentials. We were able to trace the letters UCI with this apparatus using 2.8 µm 191
polystyrene beads (15% iron) and 9.62 µm polystyrene beads. The electrodes are square shaped with 500 µm sides.
Figure 6-11: Electrode pattern used for parallel electroosmotic writing (500 µm in length).
Particle adhesion to the writing surface sometimes hindered movement of the particles when a field was applied. Larger particles tended to adhere more, perhaps due to gravity or the increase in the contact area. (106 µm glass beads were too heavy or too adhesive to the surface to be moved in a controllable manner, 9.62 µm polystyrene beads posed less a problem, and 2.8 µm polystyrene beads (15% iron) were easily manipulated.)
Experiments were also performed in a 10 mM Histidine buffer solution with a pH of 6.97. Hydrolysis and Joule heating due to the high conductivity of the solution hindered effective manipulation of the particles.
192
6.1.8. Conclusion
In this chapter, a novel lithography method using particles as transducers to trace patterns on surfaces is proposed. The two major technical issues associated with this type of lithography particle-substrate compatibility and electrokinetic nanomanipulation. We have proposed two particle-based lithography systems: particle-based lithography using dielectrophoretic traps and parallel electroosmotic writing, based respectively on dielectrophoresis and electroosmosis. The effects of miniaturizing of the particle and the electrodes on expected lithography performance were examined. Dielectrophoresis and electroosmosis scale favorably when scaling down electrode geometries and particle diameters simultaneously. The limitations in the precision of particle-based lithography systems come from Brownian motion induced errors and surface adhesion problems. Particle-substrate sensitivity along with Joule heating and hydrolysis, the latter both due to increase in field strength, are factors limiting the miniaturization of the proposed systems. Smaller particle diameters are advantageous in creating small traces, but the energy a moving particle emits diminishes with decreasing particle diameter. Thus, when the moving light-emitting particle is miniaturized, it becomes more difficult to sensitize the photosensitive substrate. Amplified sensitive surfaces that require a much smaller exposure dose such as silver halide films are needed to create small feature sizes using submicron particles. We have found that photographic film can be used as a rapid and inexpensive tool for detecting movement and location of weak luminescent particles.
193
Validation of nanomanipulation using planar and 3D electrode geometries was demonstrated using C-MEMS electrodes.
6.2.
Suggested Future Work
The work described in this dissertation can be a starting point for multiple investigations that go deeper into the potential applications of the newly introduced microfabrication techniques.
CTAB structures as deposited can be used as a sacrificial material for the formation of thick porous layers of carbon. It is suggested that a sequential deposition of CTAB and PPy can form the basis of fabrication of new porous carbon materials with controllable porosity. More research needs to be done to further validate this technique.
The stretching flow technique can be adapted into a serial method for the fabrication of suspended wires at a very specific location by using a coaxial needle that can deliver the SP SU-8 at the exact location in a precise co-flow of Isopropanol. The technique although serial in nature, can be easily parallelized for mass fabrication in a fashion similar to IBM millipede [93].
Further analysis and material characterization needs to be performed on the fast ramp pyrolysis and the associated pyrolysis under pressure to ascertain the reason for the enhanced lithium intercalation properties of such a material. 194
References 1.
2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.
Hebert, N.E., B. Snyder, R.L. McCreery, W.G. Kuhr, and S.A. Brazill, Performance of pyrolyzed photoresist carbon films in a microchip capillary electrophoresis device with sinusoidal voltammetric detection. Analytical Chemistry, 2003. 75(16): p. 4265-4271. Sun, D.Z., L.D. Zhu, and G.Y. Zhu, Glassy carbon ceramic composite electrodes. Analytica Chimica Acta, 2006. 564(2): p. 243-247. EinEli, Y. and V.R. Koch, Chemical oxidation: A route to enhanced capacity in Li-ion graphite anodes. Journal of the Electrochemical Society, 1997. 144(9): p. 2968-2973. Schueller, O.J.A., S.T. Brittain, C. Marzolin, and G.M. Whitesides, Fabrication and characterization of glassy carbon MEMS. Chemistry of Materials, 1997. 9(6): p. 1399-1406. Wang, C.L., G.Y. Jia, L.H. Taherabadi, and M.J. Madou, A novel method for the fabrication of high-aspect ratio C-MEMS structures. Journal of Microelectromechanical Systems, 2005. 14(2): p. 348-358. MicroChem, M., NaNo SU-8 data sheet. 2008. MicroChem, M., Nano SU-8 MSDS. 2008. Salamone, J.C., Polymer Materials Encylopedia. 1996, CRC. p. 1183. Mali, P., A. Sarkar, and R. Lal, Facile fabrication of microfluidic systems using electron beam lithography. Lab on a Chip, 2006. 6(2): p. 310-315. Kroto, H.W., J.R. Heath, S.C. Obrien, R.F. Curl, and R.E. Smalley, C-60 Buckminsterfullerene. Nature, 1985. 318(6042): p. 162-163. Iijima, S., Helical Microtubules of Graphitic Carbon. Nature, 1991. 354(6348): p. 56-58. Mandelbr.B, How Long Is Coast of Britain - Statistical Self-Similarity and Fractional Dimension. Science, 1967. 156(3775): p. 636-&. Edwards, B.C., A hoist to the heavens. Ieee Spectrum, 2005. 42(8): p. 36-41. Salimi, A., R. Hallaj, and G.R. Khayatian, Amperometric detection of morphine at preheated glassy carbon electrode modified with multiwall carbon nanotubes. Electroanalysis, 2005. 17(10): p. 873-879. Qi, H.L. and C.X. Zhang, Simultaneous determination of hydroquinone and catechol at a glassy carbon electrode modified with multiwall carbon nanotubes. Electroanalysis, 2005. 17(10): p. 832-838. Guan, W.J., Y. Li, Y.Q. Chen, X.B. Zhang, and G.Q. Hu, Glucose biosensor based on multi-wall carbon nanotubes and screen printed carbon electrodes. Biosensors & Bioelectronics, 2005. 21(3): p. 508-512. 195
17. 18.
19. 20. 21. 22. 23.
24. 25. 26. 27.
28. 29.
30.
Chen, R.S., W.H. Huang, H. Tong, Z.L. Wang, and J.K. Cheng, Carbon fiber nanoelectrodes modified by single-walled carbon nanotubes. Analytical Chemistry, 2003. 75(22): p. 6341-6345. Gao, B., C. Bower, J.D. Lorentzen, L. Fleming, A. Kleinhammes, X.P. Tang, L.E. McNeil, Y. Wu, and O. Zhou, Enhanced saturation lithium composition in ballmilled single-walled carbon nanotubes. Chemical Physics Letters, 2000. 327(12): p. 69-75. Park, B.Y., R. Zaouk, C.L. Wang, and M.J. Madou, A case for fractal electrodes in electrochemical applications. Journal of the Electrochemical Society, 2007. 154(2): p. P1-P5. Madou, M.J., Fundamentals of microfabrication : the science of miniaturization. 2nd ed. 2002: CRC Press. Carter, J., A. Cowen, B. Hardy , R. Mahadevan, M. Stonefield, and S. Wilcenski, PolyMUMPs Design Handbook rev.11. 2005, MEMSCAP. Park Benjamin, Y., R. Zaouk, and J. Madou Marc, Fabrication of microelectrodes using the lift-off technique. Methods in molecular biology (Clifton, N.J.), 2006. 321: p. 23-6. Ezkerra, A., L.J. Fernandez, K. Mayora, and J.M. Ruano-Lopez, Fabrication of SU-8 free-standing structures embedded in microchannels for microfluidic control. Journal of Micromechanics and Microengineering, 2007. 17(11): p. 22642271. Tuomikoski, S. and S. Franssila, Free-standing SU-8 microfluidic chips by adhesive bonding and release etching. Sensors and Actuators a-Physical, 2005. 120(2): p. 408-415. Linder, V., B.D. Gates, D. Ryan, B.A. Parviz, and G.M. Whitesides, Watersoluble sacrificial layers for surface micromachining. Small, 2005. 1(7): p. 730736. Huang, H. and C. Fu, Different fabrication methods of out-of-plane polymer hollow needle arrays and their variations. Journal of Micromechanics and Microengineering, 2007. 17(2): p. 393-402. Park, B.Y., L. Taherabadi, C. Wang, J. Zoval, and M.J. Madou, Electrical Properties and Shrinkage of Carbonized Photoresist Films and the Implications for Carbon Microelectromechanical Systems Devices in Conductive Media. Journal of the Electrochemical Society, 2005. 152(12): p. J136-J143. Lochab, J. and V.R. Singh, Acoustic behaviour of plastics for medical applications. Indian Journal of Pure & Applied Physics, 2004. 42(8): p. 595-599. Agirregabiria, M., F.J. Blanco, J. Berganzo, M.T. Arroyo, A. Fullaondo, K. Mayora, and J.M. Ruano-Lopez, Fabrication of SU-8 multilayer microstructures based on successive CMOS compatible adhesive bonding and releasing steps. Lab on a Chip, 2005. 5(5): p. 545-552. Cheng, M.C., A.P. Gadre, J.A. Garra, A.J. Nijdam, C. Luo, T.W. Schneider, R.C. White, J.F. Currie, and M. Paranjape, Dry release of polymer structures with antisticking layer. Journal of Vacuum Science & Technology A, 2004. 22(3): p. 837841. 196
31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47.
Wong, S.S., E. Joselevich, A.T. Woolley, C.L. Cheung, and C.M. Lieber, Covalently functionalized nanotubes as nanometre-sized probes in chemistry and biology. Nature, 1998. 394(6688): p. 52-55. Yu, Z., S.D. Li, and P.J. Burke, Synthesis of aligned arrays of millimeter long, straight single-walled carbon nanotubes. Chemistry of Materials, 2004. 16(18): p. 3414-3416. Li, S.D., Z. Yu, C. Rutherglen, and P.J. Burke, Electrical properties of 0.4 cm long single-walled carbon nanotubes. Nano Letters, 2004. 4(10): p. 2003-2007. Malladi, K., C.L. Wang, and M. Madou, Fabrication of suspended carbon microstructures by e-beam writer and pyrolysis. Carbon, 2006. 44(13): p. 26022607. del Campo, A. and C. Greiner, SU-8: a photoresist for high-aspect-ratio and 3D submicron lithography. Journal of Micromechanics and Microengineering, 2007. 17(6): p. R81-R95. Christian , A., Characterisation and optimisation of the variable frequency microwave technique and its application to microfabrication. 2007, Swinburne University of Technology. Russell, D.J., Controlling photoimaging of epoxies by blending to a specific molecular weight. J. Appl. Polym. Sci. FIELD Full Journal Title:Journal of Applied Polymer Science, 1990. 41(7-8): p. 1753-64. Wang, C.L. and M. Madou, From MEMS to NEMS with carbon. Biosensors & Bioelectronics, 2005. 20(10): p. 2181-2187. Kroger, R., S. Berg, A. Delgado, and H.J. Rath, Stretching Behavior of Large Polymeric and Newtonian Liquid Bridges in Plateau Simulation. Journal of NonNewtonian Fluid Mechanics, 1992. 45(3): p. 385-400. Perkins, T.T., D.E. Smith, and S. Chu, Single polymer dynamics in an elongational flow. Science, 1997. 276(5321): p. 2016-2021. Hexion , T., EPON Resin SU-8 MSDS. 2008. Mortimer, R.J., A.L. Dyer, and J.R. Reynolds, Electrochromic organic and polymeric materials for display applications. Displays, 2006. 27(1): p. 2-18. Xu, H., C. Wang, C.L. Wang, J. Zoval, and M. Madou, Polymer actuator valves toward controlled drug delivery application. Biosensors & Bioelectronics, 2006. 21(11): p. 2094-2099. Guimard, N.K., N. Gomez, and C.E. Schmidt, Conducting polymers in biomedical engineering. Progress in Polymer Science, 2007. 32(8-9): p. 876-921. Teasdale, P.R. and G.G. Wallace, In-Situ Characterization of Conducting Polymers by Measuring Dynamic Contact Angles with Wilhelmys Plate Technique. Reactive Polymers, 1995. 24(3): p. 157-164. Qu, L.T. and G.Q. Shi, Hollow microstructures of polypyrrole doped by poly(styrene sulfonic acid). Journal of Polymer Science Part a-Polymer Chemistry, 2004. 42(13): p. 3170-3177. Bolto, B.A., R. McNeill, and D.E. Weiss, Electronic Conduction in Polymers .3. Electronic Properties of Polypyrrole. Australian Journal of Chemistry, 1963. 16(6): p. 1090-&. 197
48. 49. 50. 51. 52.
53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63.
West, K., L. Bay, M.M. Nielsen, Y. Velmurugu, and S. Skaarup, Electronic conductivity of polypyrrole-dodecyl benzene sulfonate complexes. Journal of Physical Chemistry B, 2004. 108(39): p. 15001-15008. Smela, E., Microfabrication of PPy microactuators and other conjugated polymer devices. Journal of Micromechanics and Microengineering, 1999. 9(1): p. 1-18. Zhang, X.T., J. Zhang, W.H. Song, and Z.F. Liu, Controllable synthesis of conducting polypyrrole nanostructures. Journal of Physical Chemistry B, 2006. 110(3): p. 1158-1165. Siegfried, M.J. and K.S. Choi, Electrochemical crystallization of cuprous oxide with systematic shape evolution. Advanced Materials, 2004. 16(19): p. 1743-+. Yuan, W.L., E.A. O'Rear, G. Cho, G.P. Funkhouser, and D.T. Glatzhofer, Thin polypyrrole films formed on mica and alumina with and without surfactant present: characterization by scanning probe and optical microscopy. Thin Solid Films, 2001. 385(1-2): p. 96-108. Wang, C.L., L. Taherabadi, G.Y. Jia, M. Madou, Y.T. Yeh, and B. Dunn, CMEMS for the manufacture of 3D microbatteries. Electrochemical and Solid State Letters, 2004. 7(11): p. A435-A438. De Jong, K.P. and J.W. Geus, Carbon nanofibers: Catalytic synthesis and applications. Catalysis Reviews-Science and Engineering, 2000. 42(4): p. 481510. Oh, S.M. and J.Y. Lee, Structures of Pyrolytic Carbon Matrices in Carbon Carbon Composites. Carbon, 1988. 26(6): p. 763-768. Ranganathan, S. and R.L. McCreery, Electroanalytical performance of carbon films with near-atomic flatness. Analytical Chemistry, 2001. 73(5): p. 893-900. Winter, M., J.O. Besenhard, M.E. Spahr, and P. Novak, Insertion electrode materials for rechargeable lithium batteries. Advanced Materials, 1998. 10(10): p. 725-763. Zheng, T., Y.H. Liu, E.W. Fuller, S. Tseng, U. Vonsacken, and J.R. Dahn, Lithium Insertion in High-Capacity Carbonaceous Materials. Journal of the Electrochemical Society, 1995. 142(8): p. 2581-2590. Zaghib, K., X. Song, A. Guerfi, R. Kostecki, and K. Kinoshita, Effect of particle morphology on lithium intercalation rates in natural graphite. Journal of Power Sources, 2003. 124(2): p. 505-512. Kwon, K., F.P. Kong, F. McLarnon, and J.W. Evans, Characterization of the SEI on a carbon film electrode by combined EQCM and spectroscopic ellipsometry. Journal of the Electrochemical Society, 2003. 150(2): p. A229-A233. Edstrom, K., M. Herstedt, and D.P. Abraham, A new look at the solid electrolyte interphase on graphite anodes in Li-ion batteries. Journal of Power Sources, 2006. 153(2): p. 380-384. Buiel, E., LITHIUM INSERTION IN HARD CARBON ANODE MATERIALS FOR LI-ION BATTERIES. 1998, Dalhousie University: Halifax, Nova Scotia. Bard, A.J. and L.R. Faulkner, Electrochemical Methods : Fundamentals and Applications. 2nd edition ed. 2001: Wiley. 198
64. 65. 66. 67. 68.
69. 70. 71. 72. 73. 74. 75.
76.
77.
Rice, R.J. and R.L. McCreery, Quantitative Relationship between ElectronTransfer Rate and Surface Microstructure of Laser-Modified GraphiteElectrodes. Analytical Chemistry, 1989. 61(15): p. 1637-1641. Fagan, D.T., I.F. Hu, and T. Kuwana, Vacuum Heat-Treatment for Activation of Glassy-Carbon Electrodes. Analytical Chemistry, 1985. 57(14): p. 2759-2763. Markovsky, B., M.D. Levi, and D. Aurbach, The basic electroanalytical behavior of practical graphite-lithium intercalation electrodes. Electrochimica Acta, 1998. 43(16-17): p. 2287-2304. Piao, T., S.M. Park, C.H. Doh, and S.I. Moon, Intercalation of lithium ions into graphite electrodes studied by AC impedance measurements. Journal of the Electrochemical Society, 1999. 146(8): p. 2794-2798. Lust, E., G. Nurk, A. Janes, M. Arulepp, P. Nigu, P. Moller, S. Kallip, and V. Sammelselg, Electrochemical properties of nanoporous carbon electrodes in various nonaqueous electrolytes. Journal of Solid State Electrochemistry, 2003. 7(2): p. 91-105. Sawai, K. and T. Ohzuku, Factors affecting rate capability of graphite electrodes for lithium-ion batteries. Journal of the Electrochemical Society, 2003. 150(6): p. A674-A678. Yu, P., B.N. Popov, J.A. Ritter, and R.E. White, Determination of the lithium ion diffusion coefficient in graphite. Journal of the Electrochemical Society, 1999. 146(1): p. 8-14. Ho, C., I.D. Raistrick, and R.A. Huggins, Application of Ac Techniques to the Study of Lithium Diffusion in Tungsten Trioxide Thin-Films. Journal of the Electrochemical Society, 1980. 127(2): p. 343-350. Long, J.W., B. Dunn, D.R. Rolison, and H.S. White, Three-dimensional battery architectures. Chemical Reviews, 2004. 104(10): p. 4463-4492. Verbrugge, M.W., D.R. Baker, and B.J. Koch, Mathematical modeling of highpower-density insertion electrodes for lithium ion batteries. Journal of Power Sources, 2002. 110(2): p. 295-309. Hart, R.W., H.S. White, B. Dunn, and D.R. Rolison, 3-D microbatteries. Electrochemistry Communications, 2003. 5(2): p. 120-123. Klassen, B., R. Aroca, M. Nazri, and G.A. Nazri, Raman spectra and transport properties of lithium perchlorate in ethylene carbonate based binary solvent systems for lithium batteries. Journal of Physical Chemistry B, 1998. 102(24): p. 4795-4801. Dudley, J.T., D.P. Wilkinson, G. Thomas, R. Levae, S. Woo, H. Blom, C. Horvath, M.W. Juzkow, B. Denis, P. Juric, P. Aghakian, and J.R. Dahn, Conductivity of Electrolytes for Rechargeable Lithium Batteries. Journal of Power Sources, 1991. 35(1): p. 59-82. Lee, S.I., U.H. Jung, Y.S. Kim, M.H. Kim, D.J. Ahn, and H.S. Chun, A study of electrochemical kinetics of lithium ion in organic electrolytes. Korean Journal of Chemical Engineering, 2002. 19(4): p. 638-644.
199
78. 79. 80. 81. 82.
83. 84. 85. 86. 87. 88. 89. 90. 91. 92.
Levinson, H.J., Overview of lithography: challenges and metrologies. AIP Conf. Proc. FIELD Full Journal Title:AIP Conference Proceedings, 2003. 683(Characterization and Metrology for ULSI Technology): p. 365-370. Switkes, M. and M. Rothschild, Immersion lithography at 157 nm. Journal of Vacuum Science & Technology B, 2001. 19(6): p. 2353-2356. Zaidi, S.H. and S.R.J. Brueck, Multiple-Exposure Interferometric Lithography. Journal of Vacuum Science & Technology B, 1993. 11(3): p. 658-666. Berger, S.D. and J.M. Gibson, New Approach to Projection-Electron Lithography with Demonstrated 0.1 Mu-M Linewidth. Applied Physics Letters, 1990. 57(2): p. 153-155. Harriott, L.R., S.D. Berger, C. Biddick, M.I. Blakey, S.W. Bowler, K. Brady, R.M. Camarda, W.F. Connelly, A. Crorken, J. Custy, R. DeMarco, R.C. Farrow, J.A. Felker, L. Fetter, R. Freeman, L. Hopkins, H.A. Huggins, C.S. Knurek, J.S. Kraus, J.A. Liddle, M. Mkrtychan, A.E. Novembre, M.L. Peabody, R.G. Tarascon, H.H. Wade, W.K. Waskiewicz, G.P. Watson, K.S. Werder, and D. Windt, The SCALPEL Proof of Concept System. Microelectronic Engineering, 1997. 35(1-4): p. 477-480. Xia, Y.N. and G.M. Whitesides, Soft lithography. Annual Review of Materials Science, 1998. 28: p. 153-184. Schaefer, D.M., R. Reifenberger, A. Patil, and R.P. Andres, Fabrication of 2Dimensional Arrays of Nanometer-Size Clusters with the Atomic-Force Microscope. Applied Physics Letters, 1995. 66(8): p. 1012-1014. Junno, T., K. Deppert, L. Montelius, and L. Samuelson, Controlled Manipulation of Nanoparticles with an Atomic-Force Microscope. Applied Physics Letters, 1995. 66(26): p. 3627-3629. Jersch, J. and K. Dickmann, Nanostructure fabrication using laser field enhancement in the near field of a scanning tunneling microscope tip. Applied Physics Letters, 1996. 68(6): p. 868-870. McCord, M.A. and R.F.W. Pease, Lithography with the Scanning Tunneling Microscope. Journal of Vacuum Science & Technology B, 1986. 4(1): p. 86-88. Betzig, E. and R.J. Chichester, Single Molecules Observed by near-Field Scanning Optical Microscopy. Science, 1993. 262(5138): p. 1422-1425. Betzig, E., J.K. Trautman, R. Wolfe, E.M. Gyorgy, P.L. Finn, M.H. Kryder, and C.H. Chang, Near-Field Magnetooptics and High-Density Data-Storage. Applied Physics Letters, 1992. 61(2): p. 142-144. Tarun, A., M.R.H. Daza, N. Hayazawa, Y. Inouye, and S. Kawata, Apertureless optical near-field fabrication using an atomic force microscope on photoresists. Applied Physics Letters, 2002. 80(18): p. 3400-3402. Piner, R.D., J. Zhu, F. Xu, S.H. Hong, and C.A. Mirkin, "Dip-pen" nanolithography. Science, 1999. 283(5402): p. 661-663. Vettiger, P., J. Brugger, M. Despont, U. Drechsler, U. Durig, W. Haberle, M. Lutwyche, H. Rothuizen, R. Stutz, R. Widmer, and G. Binnig, Ultrahigh density, high-data-rate NEMS-based AFM data storage system. Microelectronic Engineering, 1999. 46(1-4): p. 11-17. 200
93.
94. 95. 96. 97. 98. 99. 100. 101. 102. 103.
Vettiger, P., M. Despont, U. Drechsler, U. Durig, W. Haberle, M.I. Lutwyche, H.E. Rothuizen, R. Stutz, R. Widmer, and G.K. Binnig, The "Millipede" - More than one thousand tips for future AFM data storage. Ibm Journal of Research and Development, 2000. 44(3): p. 323-340. Maier, S.A., P.G. Kik, and H.A. Atwater, Optical pulse propagation in metal nanoparticle chain waveguides. Physical Review B, 2003. 67(20). Lu, Y. and S.C. Chen, Nanopatterning of a silicon surface by near-field enhanced laser irradiation. Nanotechnology, 2003. 14(5): p. 505-508. Huang, S.M., M.H. Hong, Y.F. Lu, B.S. Lukyanchuk, W.D. Song, and T.C. Chong, Pulsed-laser assisted nanopatterning of metallic layers combined with atomic force microscopy. Journal of Applied Physics, 2002. 91(5): p. 3268-3274. Watanabe, O., T. Ikawa, M. Hasegawa, M. Tsuchimori, and Y. Kawata, Nanofabrication induced by near-field exposure from a nanosecond laser pulse. Applied Physics Letters, 2001. 79(9): p. 1366-1368. Munzer, H.J., M. Mosbacher, M. Bertsch, J. Zimmermann, P. Leiderer, and J. Boneberg, Local held enhancement effects for nanostructuring of surfaces. Journal of Microscopy-Oxford, 2001. 202: p. 129-135. Romankiw, L.T., A path: From electroplating through lithographic masks in electronics to LIGA in MEMS. Electrochimica Acta, 1997. 42(20-22): p. 29853005. Komatsu, J., M. Nakano, H. Kurita, T. Takashima, S. Katsura, and A. Mizuno, Ice-water interface migration by temperature controlling for stretching of DNA molecules. Journal of biomolecular structure & dynamics, 2004. 22(3): p. 331-7. H.Morgan, N.G., AC Electrokinetics- Colloids and particles. 2003: Research Studies Press. Castellanos, A., A. Ramos, A. Gonzalez, N.G. Green, and H. Morgan, Electrohydrodynamics and dielectrophoresis in microsystems: scaling laws. Journal of Physics D-Applied Physics, 2003. 36(20): p. 2584-2597. Voldman, J., M. Toner, M.L. Gray, and M.A. Schmidt, Design and analysis of extruded quadrupolar dielectrophoretic traps. Journal of Electrostatics, 2003. 57(1): p. 69-90.
201
Appendix SU-8 Monomer
202
PGMEA 1-METHOXY-2-PROPANOL ACETATE
203
Electrochemical Cell Rabih Zaouk, UCI MAE dept. 9558k17 Polyurethane O-ring groove 1 1 16" wide, 32" deep
1 "t. 16
3
1 8"x 2"
9558k44 polyurethane 11 2" x 13 4 " 1 8 "t.
Upper Cap Drill thru
O-ring groove 1 1 16" wide, 16 " deep
Drill Thru 9559k152 PTFE 5 16 " x 1 2 " 9558k79 Polyurethane
3
32
91746A836 Thumb screw slotted 91841A005 Nuts 1 4"x 3 32 "
Lower Cap
O-ring grooves 1 1 16" wide, 16 " deep
Drill Thru
204
